Alonzo Kelly

My research agenda emerges from the struggle to develop mobile robot systems that are useful today in applications. In my case, “useful” means that the net effect on the profit and loss of the enterprise is a positive one. It fortunately happens to (still) be the case that trying to make profitable and/or cost-effective real systems presents enough fundamental challenges to keep researchers busy. Luckily, there is never a question about whether such efforts matter in the real world, because somebody patently cares enough to part with their hard-earned money to have them solved.

So, what are those problems? My personal web site contains more details but some examples are provided below, organized by subsystems. These are often developed for either structured (indoor, road) or unstructured (off-road, off-earth) environments. The overall theme is each case is a fundamental problem of mobile robot systems was inspired from a specific application but thereafter attacked as generally as possible.

• Modeling, Simulation, and Identification:
  • general kinematic models of wheeled mobile robots undergoing slip in arbitrary terrain,
  • methods to calibrate such models in real-time based on autonomy sensing.
• Position estimation and mapping:
  • systems for factory AGVs based on image mosaic or lidar maps
  • wearable inertial navigation systems for mine workers or first responders.
• Motion Control and Path Planning:
  • algorithms to invert vehicle models to produce accurate feedforward trajectories
  • methods to generate path following and obstacle avoidance trajectories in nonholonomic control.
• Architecture and Infrastructure:
  • data structures that support efficient representation of the relative motion of the robot and the environment
  • optimal real-time control approaches to resolving the often conflicting goals of local planning and planning on the scale of 100s of kilometers.
• Human Interfaces:
  • methods to allow humans to drive robots remotely based on a synthetic rendering of the vehicle in the context of its sensor measurements of its surroundings.
  • methods to permit operators to drive robots effectively based on enhanced video whether the driver is remotely situated or inside the vehicle.
• Theoretical Mobile Robotics:
  • models of error propagation in odometry and triangulation
  • methods to optimize the design of search spaces for constraint satisfaction or probability of funding a solution in fixed time.

Additional Interests

Find more detailed information on my research interests here.