Our first idea was to create a tracked robot able to climb stairs and plenty of others obstacles, to reach a good speed (around 10 or 15 km/h) while filming and transmitting data. It will decide itself if it can go through an obstacle like a slope, stairs or a hole. Recognizing people and following them, take panorama pictures, follow a GPS designate way are its objectives.
Step 1: Chassis
Step 2: Tracks
We used :
- 4 bicycles chains that a bike store gave to us (they are about 140/150 cm)
- 4 identical gearwheels that we bought near 5 € each
- aluminum rod (about 2.5 m)
- rubber (quite the same length than aluminum rod).
We attached the chains by pairs with the aluminum rod and wire and then we attached the rubber to them.
Power wheel are made of 2 gearwheels with one wooden disc between them. Be sure to have the same space between the chains at every new component you add, then you can circle your chains.
Here is our first test od the tracks !
Step 3: Turret
Step 4: Suspensions
Step 5: Sensors
To make an autonomous robot, we had to put a lot of sensors.
First, we had to determined what we had to detect, which obstacles.
- wall (impassable)
- stairs (passable)
- ditch/hole (impassable)
So we created a simulator which contain a few obstacles, and all the sensors we needed on the robot. (http://team-gs.fr/Robot%20Chenille/simulation.html)
Then, we had fix all of this sensors.
We are using ultrasonic sensors : HC-SRO4.
It's cheap and reliable.
2) Other sensors :
We added other sensors like 3-axis magnetometer (as compass), GPS module, 3 axis accelerometer and gyroscope.
These sensors are used to optimised the movements of the robot and allowed it to know where it is, and in which position.
All of these sensors were fix with plywood and 2mm screws.
Step 6: Assembling the Robot
Step 7: Programm
The robot is Remote Controlled but not autonom.
But in a near future, we will try to implement the auto-detection of obstacle.
A video transmission of the raspi-cam to our netbook.
An automatic 360° panorama.
And finnaly, we will try to add some huge features like "people follow", "go back home", and why not face recognition ?!
Step 8: Conclusion
We will post updates as soon as possible.
And some videos will come too !
UPDATE 2013/07/08 => 2 videos
Here is the video of the first move :