Goal

Using a network of RC cars modified to work autonomously, the cars will execute instructions detailing an point a car must drive to. The cars will not know the surroundings of where they are placed beforehand, but must localize themselves first. The cars will then distribute the instructions amongst themselves.

Implementation

CarPos uses a mesh network and a distributed task queue to assign tasks between the cars. The cars are equipped with multiple sensors including Lidar, IMU, compass and optical sensor, which are attached to an onboard Raspberry Pi 3. With the help of these sensors and the use of simultaneous localization and mapping (SLAM) the CarPos cars will autonomously build a map of the surroundings and learns to navigate around obstacles when driving.

Future development

The Raspberry Pi lacks computing power which hindered our efforts with image recognition methods. In the end, we couldn't include it in the end product, but it could be used in a future version. Also designing even smarter autonomous navigation is possible.