Dieter Brehm

Gradient Descent Control

As part of Olin's Quantitative Engineering Analysis (QEA) class, I developed software in a two person team to detect obstacles using LIDAR, generate a gradient field with maximums representing objects to avoid, and a robot control system to execute a path that successfully descends that gradient field.

The development of the project was a multiple week process and required interative integration of project pieces between me and my partner. We implemented every part of the system in MATLAB, with the exception of a boilerplot ROS (Robot Operating System) node running on a raspberry pi that our professor set up.

Plot of neato gradient plot of prototype gradient and robot path produced in MATLAB

©2020 Dieter Brehm. All rights reserved.