Robotic Chemotaxis and Obstacle Avoidance
Title: Robotic Chemotaxis and Obstacle Avoidance
Student: Duncan Frasch
Advisors: Dr. Kevin Nickels and Dr. Hoa Nguyen
Abstract: Biology has been a consistent inspiration to the field of robotics. Many robotic systems operate with sensors that mimic animals’ senses of vision and hearing, and to a lesser extent, the sense of smell. The biological behavior we were inspired to mimic was chemotaxis: a method of locating and moving toward sources of food and other chemical attractants. Chemotaxis is used by organisms such as the bacterium E. coli, the silkworm moth Bombyx mori, and the dung beetle Geotrupes stercorarius. In this research project, we developed an algorithm that combined, in a novel way, the plume gradient-climbing behavior of chemotaxis with the AI-inspired behavior of obstacle avoidance which allowed a small tabletop robot to seek high chemical concentrations while avoiding collisions with obstacles. First, we implemented the chemotaxis algorithm in a robotic simulation environment. In the next step, a tabletop grayscale gradient pattern (a surrogate for the chemical plume) was detected and its source (the point of maximum darkness/concentration) located by an infrared sensor on the underside a simple robot called an E-puck. The E-puck is also equipped with radially-facing infrared sensors that can detect obstacles. The chemotaxis controller was upgraded to include obstacle detection/obstacle avoidance. The final outcome was E-pucks that reliably navigated past obstacles while searching for the source of an “attractant chemical.” This robotic system capable of locating attractant sources in complex environments could have a number of possible applications, such as detecting and sourcing chemical leaks on land and underwater, locating illegal drugs and explosives like trained dogs, or even finding truffles in a forest.