LiDAR and Mechanical Update - 3/18
This past month we’ve been continuing to evaluate our mechanical systems in order to develop actuation techniques and sensor mounting strategies. We’ve also been testing LiDAR sensor output (visualized in ROS), looking into point cloud object detection algorithms, and learning how to format our data properly for those specific techniques. Additionally, our team has cleaned and re-organized our shop space and implemented new preventive safety measures to ensure work can be done on the car efficiently while ensuring a safe environment for all students.
Inside view of the motor we will be using – from Brammo Empulse motorcycle, donated by Polaris. Taken apart to inspect how mounting orientation could affect oil lubrication within the gear train.
Saved image of LiDAR output to ROS from a SICK MRS-6124 unit. This picture is of the room surrounding the sensor. There are 24 scan layers total with a horizontal FOV of 120° and vertical FOV of 15°.
Saved image of LiDAR output to ROS from a SICK MRS-6124 unit. In this view the sensor was placed on the ground in front of a set of six cones which can be seen in the center of the image.
Sample point cloud data echoed from ROS /cloud topic published by the sick_scan package (publicly available at http://wiki.ros.org/sick_scan). This data will be fed to an additional ROS node that will parse and convert the data to the right format (as needed) for our point cloud object detection algorithms.
Example of initial motor and servo testing for a RC car. PWM control code was written on a Raspberry Pi and used with the car’s existing ESC to control motor speed and servo angle (for steering). We’ll be using the RC car to experiment with basic trajectory-following operations using PID control. Our competition car will have very different actuation, but our hope is that the RC car will allow us to evaluate concept feasibility.