The HUD was made in Node-Red. I used the default HUD, that is why there are generic looking buttons for directions. In the future, I would like to make a custom controller.
@@AvivMakesRobots So far I have seen people stream video over lora. It would be nice to have a 3D printable enclosure that can attach to say a ball cap that can stream the video over lora using a micro controller and take various common lcd type screen sizes like 1.44 or larger that will adjust for how close it is to eyes. I have been looking for that kind of a springboard to start further development for myself as I have a different type of job altogether but could help. If that is something your interested in let me know and I can send you some youtube videos that I have. I feel like people are chomping at the bit for something like this. Either way.. excellent work!
I'm a fellow roboticist studying engineering at Purdue University. I must say that I'm super inspired by your work and the CAD designs along with your sketches are so simply elegant. You've earned a subscriber and I will watch more of your videos to become a better roboticist, thank you!
Great video! I loved how you showed the design process from sketch to finished product. Excellent marriage of design and engineering. I can't wait to see how this project process!
Awesome project! To improve night vision, you can also consider adding standalone infrared LED modules. It may also help to raise the camera up a bit to expand your field of view.
Very nice combination of drawings for a first impression and then CAD for actually seeing it move. In my opinion, making (even simple) drawings of stuff is a prototyping technique that is pretty underrated. You got some pretty great results as well, I like how you built your tracks.
What earned my like was detailed part list, most channels only go far as listing motor brand and brain of the robot, but not the screws and ball bearing!
Mount the camera to a fpv gimbal with gyro sensor for more stability. Then you could add head tracking so the camera turns with your head. It uses two sealed servos, so the sand shouldn't be much of an issue.
Great build and video, especially all the editing and explanation. As far as sensors, I would consider using a depth camera (like the orbbec astra embedded s for example) You would get the image stream and also a depth stream that could be used for mapping the terrain ahead in detail so it could pick a path around obstacles. The depth part works in total darkness, but the imaging part would only work in the light.
Wow, looks like a nice camera! It would be great to get something that nice, but for now I gotta see what I can do with this budget cam. I saw your video on the AVA v2. That thing is nuts... Are you a day job robotics engineer? Programmer? I am mostly a mechanical engineer and industrial designer.
@@AvivMakesRobots I am retired, used to be a software engineer in Silicon Valley. I am self-taught in robotics and later AI. I can hang with most anyone on software, but muddle my way through mechanics and electronics. I just love the hobby. BTW, it really is a great rover you built. I wish Ava had anything like that suspension. I have a rocker suspension idea that's a little like a Sherman tank. Keep up the great work!
@@SuperDroidBot Hey, just saw this. Thanks again, I am working now on the next video which involves the vision system. I am wondering what I can do with just the pi camera using a better lens. Do you know what a good computer vision challenge for this project could be using the pi 4 and the pi camera for a beginner?
@@AvivMakesRobots You could perhaps install tensorflow and one of the versions of the yolo model...so it could recognize 80 different types of objects/people/animals, etc. The yolo model would tell you the x,y location and height, wdith in the FOV so you could use that to drive towards/away from the object or some other behavior.
Great project! May be you can look for a mechanism to reduce the vibration of the camera by suspending it to a shock absorbing spring may be. Great work.
That would be a lot simpler and robust than with a 3-axis electronic gimbal... maybe that is better than software stabilization or it could aid it. Thanks for the comment!
Wonderful in every sense of the word, all together, mechanics, electricity, electronics and programming. I wonder if it is possible to add a gyroscope system to stabilize the image coming from the camera because the image goes up and down with the movement of the wheel.
Nice flat top might be ideal for solar panel(s). Could recharge the batteries out in the field. Might be an idea to the put over a semi translucent camo over top when needed.
This is awesome! must be a complicated to 3d print and assemble the chassis You could try adding a speed ratio like crawling & running, controlled via pwm, and a few shock absorbers/ gimbal for the camera could make the footage usable. If this is supposed to be a military bot it needs to provide some use, either active combat or assistance. Active combat: reconaissance and mapping(lidar etc) or direct combat, new age weapons like rail guns etc Passive: transportation (heavy drivetrain/load bearing)& evac vehicle, comms node etc
Love the project, and I'm really enjoying the videos, especially your idea to use a drawer slide for raising the camera. Did you encounter any video signal (timing) issues by introducing cable length (and therefore a propogation delay) between the camera & the RPi? Do you have any future plans to provide a "kit" of the 3D printed parts? I don’t have a 3D printer, so that's a stumbling block for me.
Hi! I did not encounter any issues, possibly because the cable length came stock with the camera and is using USB 3.0. I absolutely want to create kits for this, I am just deciding if printing the parts myself or having them made elsewhere makes sense. If I was printing them myself and hit a snag, it could cause major delays. My future videos will touch on this topic. Stay tuned!
One question: I skipped through some parts of the video so I am not sure where the front is. Is it on the side with the camera? In the video it was shown driving in both directions. If the front is on the side with the camera I would mount the suspension facing in the other direction because this type of suspension works better in one way than the other. (But I am not sure if this is a big deal on such a small scale) I would also consider a stabilizer for the camera because it is shaking quite a bit and I would mount the camera in an elevated position instead of inside the hull for 360° vision.
Okay one last question, do you have the stl files for this exact design available to download on your site for that fee? I saw the other design but not this one available.
@@AvivMakesRobots yeah that will be cool if you can post a link under this comment are something, I would only need a few of your stl's to use in fusion.
Great project, however there are a few things you might want to adjust. I see that you're using L298N drivers, which are pretty easy to use, but are pretty limited in the power they can handle, with something like 1.5A max and can turn very hot. Specially in a 3D printed vehicle that can become an issue. You might want to concider some other drivers, or (and that would be my suggestion) go with 2 ESC's and RC control. That'll give you the option to use larger motors such as a 540 or even 775 with low RPMs, but also you can use LiPo's without having to worry about overdischarging them, and is a lot more universal. Controlling the droid by WiFi is nice, however is pretty limited and you need to concider how to handle loss of signal etc. A way more easier method would be to use RC transmitter and receiver. It's easy enough to capture the receivers output with an arduino, then you can modify your signal as pleased for either driving or other functions such as lights, and then send the output signal to ESC's or whatever you want to control. As upgrade you might want to look at adding GPS. A Neo 6M module would do perfectly fine for these kind of jobs. You can then either store the location in a timeline db such as flux, either on the Pi or sent it directly ove WiFi to a database. Another suggestion would be to add sensors for autonomous driving, with an ultrasonor sensor such as an HC-SR04 and / or use a lidar such as a TFmini. There's a lot of documentation on this subject.
Hey, I uploaded a video today on how I want to make a PCB for this. I am still using WiFi because it is easy to connect to using your cell-phone. It is also easy to connect other stuff like lidar and just program it in. With an RC transmitter, it might be more complicated. All good suggestions. I am switching to a Cytron 10 amp, 2 motor chip for the motors. I am planning on having GPS right on the chip.
@@AvivMakesRobotsRC is using a PPM signal, which isn't much different from PWM. RC gives you a bit more range, but more importantly, it'll give you a very reliable signal.
@@AvivMakesRobots I'm glad that you agree! I hope you don't mind if I elaborate... Part of the issue is that with the camera and illumination close to the ground there is a very bright patch on the ground close to the camera which the camera is auto exposing for which causes the darker background to be underexposed. With the IR lamps close to the ground the light intensity falls off fairly sharply so it is comparatively dark not far from the tank but the camera is setting it's exposure for the very bright light on the ground close to the tank - it's probably overpowering the auto exposure system. If the source of illumination and the camera are up higher the light won't be as bright right in front and the intensity of light will fall off more gradually which hopefully the camera will be able to cope with better. Also, because with the camera up higher the bright patch (which will be less intense with the IR mounter higher) will be smaller on the image/exposure sensor having less impact on the auto exposure of the camera. So the solution might be just to improve the lighting by changing the angle of illumination rather than getting a more sensitive camera or brighter IR lamps. I hope this helps.
P.S. I love the way you built the tracks - the whole build is pretty sweet actually. But the track build is instructive for anyone who wants robust 3D printed tracks!
Thanks! It does. I am looking now at 5mm infrared dip LEDs of different beam angles. It needs to hit the ground maybe 6 feet away from the robot I would say. Does that distance sound right? The angle of the LED bulb itself would be angled, calculated to hit the ground at that distance away based on whatever beam angle it is.
@@AvivMakesRobots I'm glad you found my comments helpful. I should point out that I'm speaking from my lighting experience as a photographer, not any experience specifically related to your application. I have spent a lot of time working with, setting up and observing lighting/light for many years. Also, you are lighting for a camera, so I think my experience does have some relevance here. But there may be others here who have more experience specifically related to what you are doing. Just to be clear; by "beam angle" do you mean the angle of coverage that the beam emits? I am answering your question in that context. As long as the beam is not too narrow/concentrated I don't think the beam is as important as how it is mounted because that will affect how the camera reacts and what you see. Having both the camera and the lights elevate should help, as I suggested before. But if you think your beam is too narrow I'd definitely consider a wider one. But you will still get a hot spot near to the vehicle if it is not mounted high enough. However, there are three tests that I can think of that you can do before you decide on new LEDs: 1) Temporarily mount the LEDs above the camera and possibly even back away from the front of the vehicle. 2) With the LEDs mounted higher, place a diffuser of some sort in front of the LEDs. This will scatter the beam giving a wider angle of illumination. Some possible diffusers are: A piece of tissue paper mounted on a small frame - I bet you could print a nice little frame for one, a small lens, crinkled cellophane. 3) Add to that a small screen/baffle, in front of, and below the LEDs to shade the area immediately in front of the vehicle. Adjust it to different positions to see if there is a sweet spot. Test each change sequentially. You could go straight to step 2 & 3 with the lights in their current position but it might not work. I suspect that the light source has to be further away from the camera so the mods don't interfere with the camera's exposure or create flare. Leaving the camera where it is will be OK. These changes should give you an idea of whether my original suggestions would help. Then you can consider any design and equipment changes.
With the amount of dust and sand building up on that lens it won’t be long before it is damaged. Also the video won’t be very useful if it is covered in sand. There is a reason most robots have the camera up away from the ground. Also the night vision mode may work better higher up as well, at the moment you are essentially trying to illuminate and view a plane in a bad orientation for the camera and IR lights. Having less of an angle between the ground and the camera would allow it to be illuminated better which should make the camera pick it up better too.
Very cool Project! I would only suggest to take brushless 3 phase BLDC Motors (I mean the ones with the three Lines, which are also used in Quadrocopters) for the Tank, they have at slower velocity more Power and achieves a higher final speed.
That is alotta parts!... in bike chain.. and you can sub-out parts without the 3d printer.. But cutting steel will make blista on ya fingers...Great wirk...
Excellent! The suspension "bogies" or "idlers" are mounted on angled arms. Was there a noticeable difference in the way they performed when moving forward vs reverse? (Assuming the camera is at the front). Thanks for the video.
Not much of a noticable difference, but the "backwards" way is the correct way to angle them. In the next version, the camera will be mounted on the other side to allow for this.
In case you are curious and didn't know so is this type of suspension called Christy suspension. It's well known for high cross country speed and was used in T-34 during WW2.
What if I told you I didn't know anything about coding when I started? Be jealous! And use that jealousy to build something awesome. I am waiting to see it.
I would move the camera to a head assembly that can be raised and lower for improved visibility. On indy race car camera's there was a plastic film roller so when ever it got dirty it was rotated thereby protecting the camera lens and keeping a clear view out the camera.
I gotcha. I solved this in my latest video with the acrylic shield! Also made a camera arm in the past to extend and lower the lens. Shield video here: th-cam.com/video/_8mCamJgqvg/w-d-xo.html
Four 3.7 in series, prefer to wireless pico or raspi zero 2 H (ir wireless) but no code for sharing was provided here. ESP32 wireless with 2 motor controller can be used here. Raspberry 4 requires a lot maintainance like the mars helicopter requires updating once it starts to use on Mars.
Very nice.. i can see a beach crusier outta it.. and the most expensive cooler carrier....It's an awsome build... build something with what you can found spar parts....It was great video...
You can add a 9DOF sensor, GPS and some obstacle sensor with IR don't use the ultrasonic. So you can plan autonomous routes. You can run OpenCV Python with the camera and identify objects and even people faces.
To be honest your project is fantastic but I have 3 questions How did build the body ,what kind of shock absorber do you use and what is that you bolted on the chains
Body is completely FDM 3D printed, designed in Fusion 360 CAD. I used 1:24th scale sumo RED24021 shock absorbers and I bolted horizontal plastic pieces to the chain in order to create a single belt.
How do you learn to sketch and express your ideas in drawing and design like that? That's by far the biggest problem I face every time I decide to get started on new projects?
You can start off by just doing side views of stuff, but to really sketch in 3D, you need to break into understanding "2 point perspective" and how things look in space. Lots of practice with industrial design sketching
Thank you! Someone was telling me to make it shorter so it would fit on their print bed. Longer could make it have more stability? Wider rockers could.
Can you sell plans and the files for 3D printing? A step by step how to would be great. I'm 63 and I'm learning this for the first time so I'm hoping you'll do this. Here's something, make it autonomous. If it detects a person, it will follow them around like a puppy. I have a research home project I want to use this for (paranormal study). Thanks
Hey, I can update you about the STLs in the future. The code is available for the Pi through Node Red. Check out: hackaday.io/project/181648-desert-eye-military-surveillance-robot in the downloads section for part of the code. The rest of the code I still want to debug.
Awesome build quality. I would put a 3g/4g dongle on it and make it connect home base with a vpn over the internet. That way you could command it from your home, while it being somewhere else. Which is very cool. Also, I would use a gaming controller to control it. PS4 controllers are amazing for that. Wireless. You could make it work in 2 ways. Wireless mode through bluetooth on the pi directly, or connect it to your computer and do it remotely. Its quite easy to put it all together. You could drive it watching a browser window looking through it's camera, and running a simple script which reads input from ps4 controller and sends it to the pi to be executed. If you push left controller forwards, it sends for whatever amount of time you press it the numerical value of how much you pressed it. As it is, controller gives you the exact amount from 100 (being totally pushed forward), 0, when its centered, and -100 when its totally pulled. You can use exactly those values (maybe with a reducer x0.8 or x0.5 or something like that) as input for the motor. Left joystick left motor, right joystick right motor. Its perfect. I can come up with easy ways for server/client communications too, just say so :)
I was considering eventually sketching and designing a custom handheld housing with some joystick knobs for controlling speed/ direction. I am just not sure how to work those into node red yet.
@@AvivMakesRobots The advantage of a ps4 controller is that its already completely made. Networking, or wire. Battery. Joysticks. Buttons for functions. Its all already there. Node red?
@@AvivMakesRobots You don't need node red. I would work around a web server. Easiest thing to setup. You can transmit the video from the pi, directly into your computer's browser. And build on top of that. You could have a separate script on the server, which starts the motor at the speed required by client. for instance left.php?speed=50. and you can put buttons on main page, when you click this button it sends a request to left.php to start the engine at 50 speed. alternatively, you can run a script on your computer, which reads the inputs from the joystick (easy to google the code and can run any language u like) and then sends those inputs when they change by requesting the left.php?speed=+/-100 or right.php?speed=+-100. and when joysticks are centered it stops it. push one forward, one backwards and it rotates in place. push both it will go forwards. pull both, it will drive backwards. You can make even fancier stuff, but this would be the absolute simplest i think. Now as far as what to do in that left/right php or py or perl... that depends more on the type of controller you have there. but i'm sure you can find something easy, just to set the speed and run continously until you get a command otherwise.
Can we write these codes in C language? What is the simulation application that you will use? What about the application in which the used pieces were designed?
Could you also put in a thermal vision so you would have 3 vision system's however you would need to figure out which thermal vision you want unless you have all of them especially if you have a telescopic lens for seeing at least miles away for effective seeing. And adding a weapon system.
Amazing work, man! You are very talented! I wonder if the lenses won't be obstructed with dirt... maybe a higher mount for the camera? not sure, it would expose your vehicle. You talked about making it autonomous... When I was modeling an autonomous airplane, I was going to use a cheap flight controller with GPS they sell in alixpress... maybe it would be suitable for your project as well, wit hthe right coding. Great project, man!
Battery temperature sensors to measure battery temp and release power according to temps hence saving the batteries longeitivity and safety from a possible fire hazard, this is sort of what Tesla's do in extreme temp when battery is too hot vehicle limits the speed to hence protect battery
Best way is to come up with a project you want to build and just figure it out as you go. Doing tutorials can only get you so far. I believe in self-directed work.
The intent and process of the design was very easy to understand and was very helpful. In the future, I would also like to create a robot car like this. Right now, I am designing and building a small caterpillar car that is powered by radio control. It's a toy. Designing a caterpillar is difficult but interesting.😀
LIDAR and a little turret with a laser pointer. It's just necessary. Maybe there could be a mounting bracket for it on the back and it could be detachable. :)
"Zoe the Robot" thought that this is a GREAT video. (I think she's got the HOTS for your "Military Robot") Zoe is now insisting that I build her a tracked little brother.
Thanks!! It was a lot. Mostly because of that suspension issue I showed and a bad 3D printer. It is hard to estimate in hours, but it can take 3 months if not more.
Thanks AVIV This gives me the confidence for building a full version of a tank The only thing that I don’t have is a 3D Printer 🖨️ But I can think another way to build it from other sources of materials Take my Idea 💡 The two motors that I’m using is from two E-200 Razor scooters 🛴 with the wheels 🛞 and the battery will be a 20 V Dewalt lithium battery 🔋 Lol 😂 my tank will go over 30 MPH I’m using a 30 Amp Cytron Motor Driver with some Breaker switches for protection 😂 I already built a prototype but wasn’t sure how to build a tank version They only run on scooters rubber air tires. But your video helps me a a lot Now I have to find some Bicycle Chains and a good size sprockets I have whole bunch of vinyl planks for the foot padding. I’m about 55 % getting it done ✅ Wish me luck 😂
Video Gimbal -> Can you Rig a Smooth Q2? Directional Stabilization Tripod Camera Mount -> 360 Scanning w an Iphone with Lidar.... Could that signal be relayed back via Bluetooth? Top Mounted 360 Camera for Situation Awareness?
I could, but imagine the unforgiving heat, sand particles, UV radiation and wind of the desert environment... all together could be difficult on a delicate and well tuned system like a Q2. ...But if it was packaged well and designed with robustness in mind, it could work very well.
Where is the link to the class he talks about so that we can replicate this on CAD? I looked all over the description and I couldn’t find it. And link will be tremendously appreciated.
The track stillness - illusion at least partially comes from the speed of the tracks matching the framerate of the camera viewing them. You can try adjusting the framerate to offset the track speeds, but the issue didn't present itself in later clips... so you're fine. Just something to keep in mind.
well done Aviv! that was awesome. I can tell you put a lot to that video. I think that it will be better if you actually film yourself talking instead of lip syncing. You earn my subscription!
I agree with that. I have not had problems with the belt coming off, but I think a fixed tensioner may be necessary for very, very rough terrain. Thing is, the bot is made for the rolling hills of a sandy desert and may not need to contend with rocks and sticks that will squish the suspension.
Cool! It is a good kit, but I will be coming out with an official kit for this design and people will be able to re-design and build this robot as a community.
I had considered an air pump, but was consoled by a fellow engineer that a brush could work better. The brush could be covered in sand particles in the air, but a strong air pump hose could not be harmed. The air pump would be inside the tank body and would extend to the lens with the hose.
I hope to create a printed circuit board to clean up the electronics and possibly get this thing working with ROS. That would constitute an improvement on the software side.
Cool! Only one question: Is sketching important for an Engineer? You mentioned some great sketches at the beginning... and what's called this type of sketching?
@@AvivMakesRobots Another question please, if I may: how much in total costs this robot will all its parts? Honestly, I am going to try doing something like that, and it'll be my first experience... or can you suggest something easier to start with?
I woukd suggest starting small. Do you have a 3D printer? You will need one. Get that, and spend money just on a few motors, batteries, arduino or Raspberry pi and an antenna. From there, attqch wheels you 3d print and build out a housing then learn to code your invention. Let me know your progress! I'd like to see what you come up with!
Like this project? Please give it a vote in the Make 2022 contest! make.co/contestants/desert-eye-military-surveillance-robot/
Ğ
Nice work! Did you make or buy your HUD? Can you provide some more detail on the interfacing?
Show! Fantastic.
The HUD was made in Node-Red. I used the default HUD, that is why there are generic looking buttons for directions. In the future, I would like to make a custom controller.
@@AvivMakesRobots So far I have seen people stream video over lora. It would be nice to have a 3D printable enclosure that can attach to say a ball cap that can stream the video over lora using a micro controller and take various common lcd type screen sizes like 1.44 or larger that will adjust for how close it is to eyes. I have been looking for that kind of a springboard to start further development for myself as I have a different type of job altogether but could help. If that is something your interested in let me know and I can send you some youtube videos that I have. I feel like people are chomping at the bit for something like this. Either way.. excellent work!
I'm a fellow roboticist studying engineering at Purdue University. I must say that I'm super inspired by your work and the CAD designs along with your sketches are so simply elegant. You've earned a subscriber and I will watch more of your videos to become a better roboticist, thank you!
Thank you! More coming soon
Great video! I loved how you showed the design process from sketch to finished product. Excellent marriage of design and engineering. I can't wait to see how this project process!
Awesome project! To improve night vision, you can also consider adding standalone infrared LED modules. It may also help to raise the camera up a bit to expand your field of view.
WTF Is this production quality?!! You deserve more sub. What an awesome build!
Hey, thank you! :)
I agree
This is the first time I see roller chains used to make treads and I absolutely love it. Great work.
Awesome! Next version will be even better! Stay tuned...
That exterior design is awesome: sleek, utile, and simple. Great work!
Thank you!
Very nice combination of drawings for a first impression and then CAD for actually seeing it move. In my opinion, making (even simple) drawings of stuff is a prototyping technique that is pretty underrated.
You got some pretty great results as well, I like how you built your tracks.
Thank you! The next tracks in 2.0 will be even better!
I loved how you draw in paper sketchs and more detailed drawings as well.
Paper sketches are underrated ;)
Dude.. I am blown away.. High Five when I see you. Totally off the charts. Wow....
One of the best RC Tank builds I've ever seen
Thanks for the support!
What earned my like was detailed part list, most channels only go far as listing motor brand and brain of the robot, but not the screws and ball bearing!
I went all the way, lol.
Really love the way you explained the process step by step with all the methods you used...
Thank you! I hope to go into more detail on each of these steps in future videos.
Mount the camera to a fpv gimbal with gyro sensor for more stability. Then you could add head tracking so the camera turns with your head. It uses two sealed servos, so the sand shouldn't be much of an issue.
Very cool idea! The next robot I make will have a more advanced computer vision component.
Great build and video, especially all the editing and explanation. As far as sensors, I would consider using a depth camera (like the orbbec astra embedded s for example) You would get the image stream and also a depth stream that could be used for mapping the terrain ahead in detail so it could pick a path around obstacles. The depth part works in total darkness, but the imaging part would only work in the light.
Wow, looks like a nice camera! It would be great to get something that nice, but for now I gotta see what I can do with this budget cam.
I saw your video on the AVA v2. That thing is nuts... Are you a day job robotics engineer? Programmer? I am mostly a mechanical engineer and industrial designer.
@@AvivMakesRobots I am retired, used to be a software engineer in Silicon Valley. I am self-taught in robotics and later AI. I can hang with most anyone on software, but muddle my way through mechanics and electronics. I just love the hobby. BTW, it really is a great rover you built. I wish Ava had anything like that suspension. I have a rocker suspension idea that's a little like a Sherman tank. Keep up the great work!
@@SuperDroidBot Hey, just saw this. Thanks again, I am working now on the next video which involves the vision system. I am wondering what I can do with just the pi camera using a better lens. Do you know what a good computer vision challenge for this project could be using the pi 4 and the pi camera for a beginner?
@@AvivMakesRobots You could perhaps install tensorflow and one of the versions of the yolo model...so it could recognize 80 different types of objects/people/animals, etc. The yolo model would tell you the x,y location and height, wdith in the FOV so you could use that to drive towards/away from the object or some other behavior.
I will look into this. Thank you.
Man this blows most of the raspberry pi videos out of the water. Amazing, dude.
Thanks!
Great project! May be you can look for a mechanism to reduce the vibration of the camera by suspending it to a shock absorbing spring may be. Great work.
That would be a lot simpler and robust than with a 3-axis electronic gimbal... maybe that is better than software stabilization or it could aid it. Thanks for the comment!
Wonderful in every sense of the word, all together, mechanics, electricity, electronics and programming. I wonder if it is possible to add a gyroscope system to stabilize the image coming from the camera because the image goes up and down with the movement of the wheel.
It is coming! Stay tuned...
Nice flat top might be ideal for solar panel(s). Could recharge the batteries out in the field. Might be an idea to the put over a semi translucent camo over top when needed.
Definitely considering a solar panel! That would be very cool. Maybe screen printed right on the panel.
Very cool!
This is awesome! must be a complicated to 3d print and assemble the chassis
You could try adding a speed ratio like crawling & running, controlled via pwm, and a few shock absorbers/ gimbal for the camera could make the footage usable.
If this is supposed to be a military bot it needs to provide some use, either active combat or assistance.
Active combat: reconaissance and mapping(lidar etc) or direct combat, new age weapons like rail guns etc
Passive: transportation (heavy drivetrain/load bearing)& evac vehicle, comms node etc
And since this is a raspberry Pi you should probably get a GSM hat for it
True! It will get some upgrades soon 😊
GSM hat sounds like a great idea. Wiuld that allow for 4G?
@@AvivMakesRobotsyes there are both 2G and 4G/LTE hats available
I looked into it, but the cost was bothering me. I might still consider this, though.
Love the project, and I'm really enjoying the videos, especially your idea to use a drawer slide for raising the camera. Did you encounter any video signal (timing) issues by introducing cable length (and therefore a propogation delay) between the camera & the RPi?
Do you have any future plans to provide a "kit" of the 3D printed parts? I don’t have a 3D printer, so that's a stumbling block for me.
Hi! I did not encounter any issues, possibly because the cable length came stock with the camera and is using USB 3.0. I absolutely want to create kits for this, I am just deciding if printing the parts myself or having them made elsewhere makes sense. If I was printing them myself and hit a snag, it could cause major delays. My future videos will touch on this topic. Stay tuned!
One question: I skipped through some parts of the video so I am not sure where the front is. Is it on the side with the camera? In the video it was shown driving in both directions.
If the front is on the side with the camera I would mount the suspension facing in the other direction because this type of suspension works better in one way than the other. (But I am not sure if this is a big deal on such a small scale) I would also consider a stabilizer for the camera because it is shaking quite a bit and I would mount the camera in an elevated position instead of inside the hull for 360° vision.
Definitely, and I agree about the direction. The next version will have a camera mounted on the suspension front of the robot which is the real front.
the best rc diy design an concept i ever seen in youtube 😎👍, keep working bro 👍
...And we're just gettin' started 🔥.
Okay one last question, do you have the stl files for this exact design available to download on your site for that fee? I saw the other design but not this one available.
I don't have this one, but should probably make it available. I can do it if you want.
@@AvivMakesRobots yeah that will be cool if you can post a link under this comment are something, I would only need a few of your stl's to use in fusion.
Great project, however there are a few things you might want to adjust. I see that you're using L298N drivers, which are pretty easy to use, but are pretty limited in the power they can handle, with something like 1.5A max and can turn very hot. Specially in a 3D printed vehicle that can become an issue. You might want to concider some other drivers, or (and that would be my suggestion) go with 2 ESC's and RC control. That'll give you the option to use larger motors such as a 540 or even 775 with low RPMs, but also you can use LiPo's without having to worry about overdischarging them, and is a lot more universal. Controlling the droid by WiFi is nice, however is pretty limited and you need to concider how to handle loss of signal etc. A way more easier method would be to use RC transmitter and receiver. It's easy enough to capture the receivers output with an arduino, then you can modify your signal as pleased for either driving or other functions such as lights, and then send the output signal to ESC's or whatever you want to control.
As upgrade you might want to look at adding GPS. A Neo 6M module would do perfectly fine for these kind of jobs. You can then either store the location in a timeline db such as flux, either on the Pi or sent it directly ove WiFi to a database.
Another suggestion would be to add sensors for autonomous driving, with an ultrasonor sensor such as an HC-SR04 and / or use a lidar such as a TFmini. There's a lot of documentation on this subject.
Hey, I uploaded a video today on how I want to make a PCB for this. I am still using WiFi because it is easy to connect to using your cell-phone. It is also easy to connect other stuff like lidar and just program it in. With an RC transmitter, it might be more complicated. All good suggestions. I am switching to a Cytron 10 amp, 2 motor chip for the motors. I am planning on having GPS right on the chip.
@@AvivMakesRobotsRC is using a PPM signal, which isn't much different from PWM. RC gives you a bit more range, but more importantly, it'll give you a very reliable signal.
Raising the camera and IR onto a mount on top of the chassis would improve the IR illumination.
True. I am considering a solution much like this one.
@@AvivMakesRobots I'm glad that you agree! I hope you don't mind if I elaborate...
Part of the issue is that with the camera and illumination close to the ground there is a very bright patch on the ground close to the camera which the camera is auto exposing for which causes the darker background to be underexposed.
With the IR lamps close to the ground the light intensity falls off fairly sharply so it is comparatively dark not far from the tank but the camera is setting it's exposure for the very bright light on the ground close to the tank - it's probably overpowering the auto exposure system.
If the source of illumination and the camera are up higher the light won't be as bright right in front and the intensity of light will fall off more gradually which hopefully the camera will be able to cope with better. Also, because with the camera up higher the bright patch (which will be less intense with the IR mounter higher) will be smaller on the image/exposure sensor having less impact on the auto exposure of the camera.
So the solution might be just to improve the lighting by changing the angle of illumination rather than getting a more sensitive camera or brighter IR lamps. I hope this helps.
P.S. I love the way you built the tracks - the whole build is pretty sweet actually. But the track build is instructive for anyone who wants robust 3D printed tracks!
Thanks! It does. I am looking now at 5mm infrared dip LEDs of different beam angles. It needs to hit the ground maybe 6 feet away from the robot I would say. Does that distance sound right? The angle of the LED bulb itself would be angled, calculated to hit the ground at that distance away based on whatever beam angle it is.
@@AvivMakesRobots I'm glad you found my comments helpful.
I should point out that I'm speaking from my lighting experience as a photographer, not any experience specifically related to your application. I have spent a lot of time working with, setting up and observing lighting/light for many years. Also, you are lighting for a camera, so I think my experience does have some relevance here. But there may be others here who have more experience specifically related to what you are doing.
Just to be clear; by "beam angle" do you mean the angle of coverage that the beam emits? I am answering your question in that context.
As long as the beam is not too narrow/concentrated I don't think the beam is as important as how it is mounted because that will affect how the camera reacts and what you see. Having both the camera and the lights elevate should help, as I suggested before. But if you think your beam is too narrow I'd definitely consider a wider one. But you will still get a hot spot near to the vehicle if it is not mounted high enough.
However, there are three tests that I can think of that you can do before you decide on new LEDs:
1) Temporarily mount the LEDs above the camera and possibly even back away from the front of the vehicle.
2) With the LEDs mounted higher, place a diffuser of some sort in front of the LEDs. This will scatter the beam giving a wider angle of illumination.
Some possible diffusers are: A piece of tissue paper mounted on a small frame - I bet you could print a nice little frame for one, a small lens, crinkled cellophane.
3) Add to that a small screen/baffle, in front of, and below the LEDs to shade the area immediately in front of the vehicle. Adjust it to different positions to see if there is a sweet spot.
Test each change sequentially.
You could go straight to step 2 & 3 with the lights in their current position but it might not work. I suspect that the light source has to be further away from the camera so the mods don't interfere with the camera's exposure or create flare. Leaving the camera where it is will be OK. These changes should give you an idea of whether my original suggestions would help. Then you can consider any design and equipment changes.
With the amount of dust and sand building up on that lens it won’t be long before it is damaged. Also the video won’t be very useful if it is covered in sand. There is a reason most robots have the camera up away from the ground. Also the night vision mode may work better higher up as well, at the moment you are essentially trying to illuminate and view a plane in a bad orientation for the camera and IR lights. Having less of an angle between the ground and the camera would allow it to be illuminated better which should make the camera pick it up better too.
True! The newer version solved both these problems.
Very cool Project!
I would only suggest to take brushless 3 phase BLDC Motors (I mean the ones with the three Lines, which are also used in Quadrocopters) for the Tank, they have at slower velocity more Power and achieves a higher final speed.
Very cool. It's a good idea, but is there enough torque with no gearbox?
That is alotta parts!... in bike chain.. and you can sub-out parts without the 3d printer.. But cutting steel will make blista on ya fingers...Great wirk...
Agreed! Wayy too complex and the next version will not use it.
Excellent! The suspension "bogies" or "idlers" are mounted on angled arms. Was there a noticeable difference in the way they performed when moving forward vs reverse? (Assuming the camera is at the front). Thanks for the video.
Not much of a noticable difference, but the "backwards" way is the correct way to angle them. In the next version, the camera will be mounted on the other side to allow for this.
In case you are curious and didn't know so is this type of suspension called Christy suspension. It's well known for high cross country speed and was used in T-34 during WW2.
True! It is also a simple suspension technique vs other types.
I enjoyed your video, & you should make a chassis that if in the unit got tipped over it would still have the capabilities to keep going
It would be cool to have a solenoid that triggers to flip the robot over.
This may be a stupid question but why does the gear not seem to be moving at 7:26? Is this a trick of the light, or is there something I'm missing?
Is that the frame rate issue people keep talking about?
@@AvivMakesRobots Ah, that might be it, if the framerate is close to the rate the wheel spins.
this is insane! I LOVE IT!!
Thank you!
Very cool build and great video quality!!!
Thank you.
Wow !!! This man knows how to design , build and code … freaking jealous of you man
What if I told you I didn't know anything about coding when I started? Be jealous! And use that jealousy to build something awesome. I am waiting to see it.
@@AvivMakesRobots that’s so nice of you and you are right ! Thank you Aviv
Awesome !!
What about the dust that is gathering on the lens?
Also if you can add image stabiliser, the video feed will be more stable.
Yes, I need to find a way to add an image stabilizer.
I would move the camera to a head assembly that can be raised and lower for improved visibility. On indy race car camera's there was a plastic film roller so when ever it got dirty it was rotated thereby protecting the camera lens and keeping a clear view out the camera.
I gotcha. I solved this in my latest video with the acrylic shield! Also made a camera arm in the past to extend and lower the lens. Shield video here: th-cam.com/video/_8mCamJgqvg/w-d-xo.html
Four 3.7 in series, prefer to wireless pico or raspi zero 2 H (ir wireless) but no code for sharing was provided here. ESP32 wireless with 2 motor controller can be used here. Raspberry 4 requires a lot maintainance like the mars helicopter requires updating once it starts to use on Mars.
I have a separate video on my channel for the code I used on the motors.
Very nice.. i can see a beach crusier outta it.. and the most expensive cooler carrier....It's an awsome build... build something with what you can found spar parts....It was great video...
Haha! Thanks. Next one is coming!
Excellent engineering and very impressive production quality! I just subscribed.
Thanks! More coming!
You can add a 9DOF sensor, GPS and some obstacle sensor with IR don't use the ultrasonic.
So you can plan autonomous routes.
You can run OpenCV Python with the camera and identify objects and even people faces.
OpenCV is absolutely a next step. GPS for way-points would be really cool. I hope I have enough GPIO pins for all of this.
add a top mounted camera with a stabilizer and thermal and sell it to the military as a scout vehicle
I want to put a thermal camera on it for sure.
This is just great, awesome work man, one question, what shock absorbers have you used?
Thank you! Here is the link to the shocks (as an Amazon Associate I earn from qualifying purchases): amzn.to/2S1Acsu
To be honest your project is fantastic but I have 3 questions
How did build the body ,what kind of shock absorber do you use and what is that you bolted on the chains
Body is completely FDM 3D printed, designed in Fusion 360 CAD. I used 1:24th scale sumo RED24021 shock absorbers and I bolted horizontal plastic pieces to the chain in order to create a single belt.
Very good project.
Good modeling process.
Thank you!
How do you learn to sketch and express your ideas in drawing and design like that? That's by far the biggest problem I face every time I decide to get started on new projects?
You can start off by just doing side views of stuff, but to really sketch in 3D, you need to break into understanding "2 point perspective" and how things look in space. Lots of practice with industrial design sketching
Very very inspiring, thank you for sharing !!
You got it!
You have nice drawing skills! This chassis should be longer to have better off-road performance
Thank you! Someone was telling me to make it shorter so it would fit on their print bed. Longer could make it have more stability? Wider rockers could.
Can you sell plans and the files for 3D printing? A step by step how to would be great. I'm 63 and I'm learning this for the first time so I'm hoping you'll do this. Here's something, make it autonomous. If it detects a person, it will follow them around like a puppy. I have a research home project I want to use this for (paranormal study). Thanks
Yes, at the bottom of my new site, you can purchase the STL files for the new version. DesertEyeRobotics.com
Very cool robot! - Subscribed.
This is an excellent candidate for using ROS.
Thank you! I keep wanting to do a simulation in Gazebo, but am more interested in doing a dynamical mechanical simulation.
Wow! Awesome drawing skills!
Thank you! Maybe I will show more of that.
Hi, I would like to have a go at building this. Are the stl files available to buy? Also is the code available for the raspberry pi?
Hey, I can update you about the STLs in the future. The code is available for the Pi through Node Red. Check out: hackaday.io/project/181648-desert-eye-military-surveillance-robot in the downloads section for part of the code. The rest of the code I still want to debug.
@@AvivMakesRobots Hi Mate, Thanks for the reply
Sure. The grand plan is to create a kickstarter campaign where you can buy pre-printed pieces as part of a kit.
Awesome build quality. I would put a 3g/4g dongle on it and make it connect home base with a vpn over the internet. That way you could command it from your home, while it being somewhere else. Which is very cool. Also, I would use a gaming controller to control it. PS4 controllers are amazing for that. Wireless. You could make it work in 2 ways. Wireless mode through bluetooth on the pi directly, or connect it to your computer and do it remotely. Its quite easy to put it all together. You could drive it watching a browser window looking through it's camera, and running a simple script which reads input from ps4 controller and sends it to the pi to be executed. If you push left controller forwards, it sends for whatever amount of time you press it the numerical value of how much you pressed it. As it is, controller gives you the exact amount from 100 (being totally pushed forward), 0, when its centered, and -100 when its totally pulled. You can use exactly those values (maybe with a reducer x0.8 or x0.5 or something like that) as input for the motor. Left joystick left motor, right joystick right motor. Its perfect. I can come up with easy ways for server/client communications too, just say so :)
I was considering eventually sketching and designing a custom handheld housing with some joystick knobs for controlling speed/ direction. I am just not sure how to work those into node red yet.
@@AvivMakesRobots The advantage of a ps4 controller is that its already completely made. Networking, or wire. Battery. Joysticks. Buttons for functions. Its all already there. Node red?
@@AvivMakesRobots You don't need node red. I would work around a web server. Easiest thing to setup. You can transmit the video from the pi, directly into your computer's browser. And build on top of that. You could have a separate script on the server, which starts the motor at the speed required by client. for instance left.php?speed=50. and you can put buttons on main page, when you click this button it sends a request to left.php to start the engine at 50 speed.
alternatively, you can run a script on your computer, which reads the inputs from the joystick (easy to google the code and can run any language u like) and then sends those inputs when they change by requesting the left.php?speed=+/-100 or right.php?speed=+-100. and when joysticks are centered it stops it. push one forward, one backwards and it rotates in place. push both it will go forwards. pull both, it will drive backwards.
You can make even fancier stuff, but this would be the absolute simplest i think.
Now as far as what to do in that left/right php or py or perl... that depends more on the type of controller you have there. but i'm sure you can find something easy, just to set the speed and run continously until you get a command otherwise.
you should add a mast on your robot and mount the camera higher. good luck. also, nice quality of video! 👍🏻
Thank you! I am working on a new lens integration using this same camera sensor.
Can we write these codes in C language? What is the simulation application that you will use? What about the application in which the used pieces were designed?
I wrote mostly in Node Red. There was no simulation application. Fusion 360 was used to design the whole physical thing.
Well 😊thank you for your response👍😍.
Could you also put in a thermal vision so you would have 3 vision system's however you would need to figure out which thermal vision you want unless you have all of them especially if you have a telescopic lens for seeing at least miles away for effective seeing. And adding a weapon system.
Check out my video "robotic vision system!" I made a telescoping lens work on the robot.
Amazing work, man! You are very talented! I wonder if the lenses won't be obstructed with dirt... maybe a higher mount for the camera? not sure, it would expose your vehicle. You talked about making it autonomous... When I was modeling an autonomous airplane, I was going to use a cheap flight controller with GPS they sell in alixpress... maybe it would be suitable for your project as well, wit hthe right coding. Great project, man!
That's a good idea! I am building integration with the Raspi... 2.0 coming soon!
Battery temperature sensors to measure battery temp and release power according to temps hence saving the batteries longeitivity and safety from a possible fire hazard, this is sort of what Tesla's do in extreme temp when battery is too hot vehicle limits the speed to hence protect battery
I will remember this tip.
Sir how to learn to design in cad understand the concepts and 3d printing ?
Best way is to come up with a project you want to build and just figure it out as you go. Doing tutorials can only get you so far. I believe in self-directed work.
The intent and process of the design was very easy to understand and was very helpful. In the future, I would also like to create a robot car like this. Right now, I am designing and building a small caterpillar car that is powered by radio control. It's a toy. Designing a caterpillar is difficult but interesting.😀
Very cool!
Good job. We need more people like you. You are very smart.
Thanks, but I am no genius. I just have boundless curiosity mixed with work habits.
thank you so much for this video it has gave me soo much inspiration for my build
Great to hear!
5:10 What are those battery endplates called and where did you get them!?
I added a link to the plates in the video description under "Electrical" :)
@@AvivMakesRobots Nice! Thx!
Did you buy the tread pieces online or 3D print them because I can’t find them anywhere in your description. Great job by the way!
Thank you! I made them.
Amazing work, but the suspension should've been less angled so larger rocks wouldn't make the suspension go in the opposite direction
Thank you! The suspension arms?
@@AvivMakesRobots yes
You mean they can fold in so much that actually go past 0 degrees. I see.
LIDAR and a little turret with a laser pointer. It's just necessary. Maybe there could be a mounting bracket for it on the back and it could be detachable. :)
LIDAR is a cool idea.
"Zoe the Robot" thought that this is a GREAT video. (I think she's got the HOTS for your "Military Robot") Zoe is now insisting that I build her a tracked little brother.
Haha
it's what could be the most suitable for digging underground, Imagination
That would be awesome! I know one of the space lander robots has a drill. Honeybee robotics made a drill too, I believe.
@@AvivMakesRobots
a bicycle chain sandwiched with chainsaw chain links,, I try to imagine a machine that can go underground
Love this project!! If you had to estimate, how many hours did you spend on this project overall?
Thanks!! It was a lot. Mostly because of that suspension issue I showed and a bad 3D printer. It is hard to estimate in hours, but it can take 3 months if not more.
very nice design and superbly made video!
Thank you!
Thanks AVIV
This gives me the confidence for building a full version of a tank
The only thing that I don’t have is a 3D Printer 🖨️
But I can think another way to build it from other sources of materials
Take my Idea 💡
The two motors that I’m using is from two E-200 Razor scooters 🛴 with the wheels 🛞 and the battery will be a 20 V Dewalt lithium battery 🔋
Lol 😂 my tank will go over 30 MPH
I’m using a 30 Amp Cytron Motor Driver with some Breaker switches for protection 😂
I already built a prototype but wasn’t sure how to build a tank version
They only run on scooters rubber air tires.
But your video helps me a a lot
Now I have to find some Bicycle Chains and a good size sprockets
I have whole bunch of vinyl planks for the foot padding.
I’m about 55 % getting it done ✅
Wish me luck 😂
Awesome! I want you to make a video of this thing when you're done.
Very cool indeed pal, thanks for sharing.
You got it! And I'm just getting started .
what video card do you have running solidworks on?
Hey! This is Fusion running on a Geforce GTX970. I am sure there are video cards more suited to 3D CaDD, a Quattro perhaps.
@@AvivMakesRobots Ah! I thought it was solidworks. Indeed, solidworks works well only on quattro.
What size and type of chain did you use?
No. 25 steel roller chain! Suggest a plastic over this for the weight reduction, though!
Excellent content, and well designed
Thanks!
Video Gimbal -> Can you Rig a Smooth Q2?
Directional Stabilization
Tripod Camera Mount -> 360 Scanning w an Iphone with Lidar.... Could that signal be relayed back via Bluetooth?
Top Mounted 360 Camera for Situation Awareness?
I could, but imagine the unforgiving heat, sand particles, UV radiation and wind of the desert environment... all together could be difficult on a delicate and well tuned system like a Q2. ...But if it was packaged well and designed with robustness in mind, it could work very well.
Where is the link to the class he talks about so that we can replicate this on CAD? I looked all over the description and I couldn’t find it. And link will be tremendously appreciated.
Playlist link is at the end of the video!
First few seconds and you already got my subscription! Amazing work!
Alright!
The track stillness - illusion at least partially comes from the speed of the tracks matching the framerate of the camera viewing them. You can try adjusting the framerate to offset the track speeds, but the issue didn't present itself in later clips... so you're fine. Just something to keep in mind.
Gotcha! There was a lot to keep track of xD, but that is the fun of it.
@@AvivMakesRobots Yeah, no, great job with the whole thing though :) Super impressed.
Thank you. I have so much more I want to show off. Super excited.
I was looking for idea how to make tank tracks. Tks for your video to give me an idea sir. Good luck
Thank you! Can't wait to see what you come up with.
well done Aviv! that was awesome. I can tell you put a lot to that video. I think that it will be better if you actually film yourself talking instead of lip syncing. You earn my subscription!
Thanks for watching!
How do you control it using wifi? Any resources to learn the same...
I learned Node-Red programming.
What if you put an extra auto tensioner pulley on the top part of the track? That slack in the belt makes my skin itch for some reason.
I agree with that. I have not had problems with the belt coming off, but I think a fixed tensioner may be necessary for very, very rough terrain. Thing is, the bot is made for the rolling hills of a sandy desert and may not need to contend with rocks and sticks that will squish the suspension.
Well done, Looks nice too, I am too building one from TS100 kit
Cool! It is a good kit, but I will be coming out with an official kit for this design and people will be able to re-design and build this robot as a community.
@@AvivMakesRobots That will be awesome, the suspension on the TS100 kit is not as good as yours anyways.
7:53 be careful with some hand actions :)
Top video and great production quality
Meme me!
solution for cleaning the dust on the lenses ?
I had considered an air pump, but was consoled by a fellow engineer that a brush could work better. The brush could be covered in sand particles in the air, but a strong air pump hose could not be harmed. The air pump would be inside the tank body and would extend to the lens with the hose.
It would be much better if you add vertical stabilization for cameras module.
Exactly what I was thinking.
Cool start but still some ways to go. Are you adhering to a particular standard or do you plan on doing so in the future?
What kind of standard?
@@AvivMakesRobots An example would be some ISO standard for environmental protection, or a military standard for operation in certain environments.
I hope to create a printed circuit board to clean up the electronics and possibly get this thing working with ROS. That would constitute an improvement on the software side.
Cool! Only one question: Is sketching important for an Engineer?
You mentioned some great sketches at the beginning... and what's called this type of sketching?
Most engineers do not do much sketching past the basics. That is industrial design sketching.
@@AvivMakesRobots Another question please, if I may: how much in total costs this robot will all its parts? Honestly, I am going to try doing something like that, and it'll be my first experience... or can you suggest something easier to start with?
I woukd suggest starting small. Do you have a 3D printer? You will need one. Get that, and spend money just on a few motors, batteries, arduino or Raspberry pi and an antenna. From there, attqch wheels you 3d print and build out a housing then learn to code your invention. Let me know your progress! I'd like to see what you come up with!
You could attached thin solar plate, whatever good work 👍🏿
Thank you! If you have a solar panel suggestion, I am all ears!
Aviv, you could make it deliver and set directional charges for example - that would be a good application.
When you say directional charges, you are referring to construction? I think anything is possible once localization is achieved!
Love that project!
Thank you!
Awesome project, I like I like
Thank you!
Excelente proyecto.
Muchas gracias!
Can it go through snow and other terrains? Great video btw. Subbed : )
Thanks!
Very cool design and Plat form
Thank you!
Fusion360 is the best cad! Great vid! Next time can you make a two legged robot?
That would be sweet! I hope to perfect this robot first before moving on- maybe making a more stabilized and usable camera first.
@@AvivMakesRobots Thats great!!
Underrated asf