Projects

Projects are often done in collaboration with an industry sponsor and supported by NSERC, FRQNT, or Mitacs. Support of both industry partners and government funding agencies is graciously acknowledged. 

For a list of publications related to these projects, see google scholar.

 

Current Projects

Extension and application of the invariant extended Kalman filter on matrix Lie groups. For example, invariant forms of batch estimation, SLAM, monocular-IMU navigation, and others.
Using a laser scanner, DVL, and IMU, ascertain the position of an AUV, as well as map the surrounding underwater environment. 
Koopman operator theory for data driven modelling and control for nonlinear systems.
Research and development of state-estimation and formation control algorithms for groups of robots using UWB signals for localization. In collaboration with the Mobile Robotics and Autonomous Systems Laboratory, École Polytechnique, Montréal.
Stereo SLAM as well as dynamic object tracking for ground vehicles in invariant framework. 
Improving stability and transparency of haptic devices through novel control strategies.

Past Projects

When GPS data is poor or unavailable, estimate a train's position using an IMU only. Specifically, exploit the fact the train is constrained to a railway track with unique features (e.g., unique curvature, grade, etc.)
Research and implementation of fully autonomous feedback control strategies for industrial crane systems.
Design of high-performance control architectures for ultra-compact industrial robot arms. Reduce gearbox-induced vibration as well as increase frequency of pick-and-place operations. 
Develop a tightly-coupled Visual-Inertial-LIDAR simultaneous localization and mapping (SLAM) algorithm for unmanned aerial vehicle (UAV) partnership with ARA Robotique.
Develop a system that can test mechanical and contractile functionality of human or animal muscle tissues while exposed to drugs, biological mediators, and/or environmental factors.
Building models for haptic feedback and control in spinal interbody fusion surgical simulations.
An reinforcement-learning (RL) based method of controlling physically simulated humanoids that learns from unstructured motion-capture, emphasizing responsiveness to user input, quality, and low runtime cost for application in video-games.
Research on navigation, guidance, and control (NGC) of unconventional VTOL UAVs, focus on real-time estimation of wind velocity using standard sensors onboard a UAV. Use of machine learning techniques.
Guidance and control of multicopter landing using MPC  on matrix Lie group. 
Homography estimation and robot navigation using thermal cameras.