rc-car-ouster-lidar-based-ml-model-deployment-general-laser.png
The purpose of this project was to develop a Machine Learning model to enable an RC car to autonomously navigate a race track using an Ouster OS1 lidar sensor as the primary sensor input. The model is an end-to-end Convolutional Neural Network (CNN) that processes intensity image data from the lidar and outputs a steering command for the RC car.
This is inspired by the Udacity Self Driving Car “behavior cloning” module as well as the DIY Robocars races. This project builds upon earlier work developing a similar model using color camera images from a webcam to autonomously navigate the RC car.
Platform Overview
For this project, the base RC platform is the Exceed Blaze (Hyper Blue) vehicle. To mount the hardware, the platform was modified with the Standard Donkey Car Kit.
For sensing, has been used an Ouster OS1-64 lidar sensor. The system also has a Genius WideCam F100 webcam and an OpenMV Cam M7 mounted but they were not used for this project.
Lidar sensors traditionally output a 3D pointcloud which contains rich spatial information. However, this data is more challenging to work with compared to camera images in ML applications. The blog post, “The Lidar IS theCamera” details how the OS1 lidar is able to output fixed resolution depth images, signal images, and ambient images in real-time, all without a camera. The open source driver outputs these data layers as fixed resolution 360° panoramic frames. This results in camera-like images that can be used with the same ML models developed using a color camera. As the blog post states, the OS1 captures both the signal and ambient data in the near infrared, so the data closely resembles visible light images of the same scenes, which gives the data a natural look and a high chance that algorithms developed for cameras translate well to the data.
All processing is done on a Compulab fitlet2 with an Arduino Uno providing commands to the motors for steering and throttle via a PCA 9685 Servo Driver. The vehicle can be operated remotely with an Xbox controller.
For further information: Ouster Blog
