Freespace Segmentation on Autonomous Robots
This repository contains robot-control and wheelchair-control code used to test my research work on semantic segmentation for detection of negative obstacles. In particular, this repo includes the integration code used to test the system on a full-sized electric wheelchair (an STM32 is used as the bridge between the wheelchair's internal CAN bus and the ROS system running on a Jetson TX2, which runs the control and deep learning software).
Please see https://github.com/asinghani/freespace-nn for the core neural-network training associated with this project.