I’ve been looking for a Raspberry-based robot for a project I’m working on.
Some of the projects needs are:
- 4-6 robots controlled by a main computer. Robots report sensor readings and their position to the main computer, which controls their movements.
- Robots have the ability to localise themselves in an arena (for example, using on of the solutions mentioned here).
- General Raspberry-Pi modularity when it comes to sensors - I have a bunch of temp/humidity/sound sensors from a previous project I’d prefer using.
The GoPiGo seems like a very good robot solution, but I couldn’t find any material about the points I mentioned. Any help/pointers would be much appreciated!
To my knowledge, no one has demonstrated a GoPiGo3 with SLAM using any of the techniques you linked, or using GoPiGo3 odometry and the DI servo mounted DI distance sensor.
Of the techniques linked, multiple April Tags in the field of view of a Pi-cam 2 (wider FOV 62deg vs 53deg of v1.3 cam) offer the highest prob of success in a small space.
I don’t expect a RPi3B+ will be able to do the localization from recognizing multiple tags in an image very fast. The demo video I found used pretty large tags about every 3-4 meters if memory serves.
Just the localization problem will take considerable effort, then adding the data and command channels to the off platform processor is another challenge (unless you are planning a ROS based solution)
Keep us posted please
I think you could get to something by using a 2D lidar and a bunch of beacons positioned in all 4 corners of the arena.
Both of these (both the lidar and the network of beacons) are going to complement each other to create a final estimation of the position of the robot within the arena.
That’s my rough idea on this.