Charles C. Cossette

Charles Cossette

I am currently a Software Engineer at Zoox, where I work on the autonomous vehicle's collision avoidance system. I previously completed my Ph.D. in Robotics with the DECAR Group at McGill university, under the supervision of James Forbes and David SaussiƩ.

At Zoox, I'm currently developing algorithms that predict the motion of other agents/objects around the vehicle, and to correspondingly compute collision probabilities. During my PhD, our research focused on planning, estimation, and control algorithms for multi-robot teams. We specifically researched how to best use ultra-wideband radio to enable collaboration between robots.

Some things I've done

Drones in flight

Optimal multi-robot formations for relative pose estimation using range measurements

A paper quantifying and describing how robots can put themselves in particular formations that will provide the best range-based relative positioning accuracy.

Drone relative positioning

Relative position estimation between two UWB devices with IMUs

A paper describing a real-time algorithm for estimating the relative position between two robots, each possessing a single ultra-wideband transceiver, and a 9DOF inertial measurement unit.

Python matrix lie groups

PyMLG - Matrix Lie Groups in PyTorch, Jax, Numpy, and C++

Pure-static implementations in python for common matrix Lie groups. Their definition as pure-static classes makes usage ultra simple.

navlie

navlie - A python package for state estimation on Lie groups

A problem-agnostic state estimation library for robotics, containing a variety of ready-to-use algorithms.

rocketry

High-powered Rocketry

An old passion of mine. As part of an engineering design team, I competed in international rocketry competitions, where the challenge was to design a rocket to hit a specific altitude with precision.