Work

Twenty-five years of autonomy for vehicles that operate where there are no lane markings — unimproved roads, construction sites, dirt pits, and the terrain between them.

Technical domains

  • Off-road autonomy & navigation
  • Manned / unmanned teaming
  • Mission planning & management
  • Bandwidth-constrained teleoperation
  • Perception & obstacle detection

Neya Systems

Founder & CEO · 2009–18

As founder and CEO my job was to grow the company and set technical direction, not to write the code — the work below belongs to a great many talented people. A broader selection lives on the Neya YouTube channel; much of the rest can’t be shown.

Autonomy overview

Wheel-placement planning in heavy clutter, adaptive learning for stereo-generated world models, trail and construction-site navigation, person following, constrained teleoperation, and backing long-combination trailers.

Manned / unmanned teaming

Built with SoarTech: ground vehicles that track and follow individual soldiers, hold a position in formation, and run movement and maneuver tactics — bounding overwatch, taking point, rear guard.

Mission planning & management

Off-line plans authored by experts in a domain-specific planning language, task decomposition to break large tasks into smaller ones, market-based delegation across available robots, and execution monitoring that triggers re-planning when the world disagrees with the plan.

Applied Perception / QinetiQ North America

Director, Projects & Ops · 2003–09

I arrived as a research engineer and left running projects and operations, including two years at QinetiQ North America after the 2007 acquisition. Lead architect and principal author of the autonomy stack — sensor and platform drivers, perception, path planning — and author of dozens of winning SBIR, BAA, and commercial proposals.

FIRRE — Family of Integrated Rapid Reconnaissance Equipment

An Army Maneuver Support Center program for fixed and mobile rapid-reconnaissance equipment. I developed the autonomy for the large mobile platform.

FIRRE — specifications

Tracking
2–3 cm precision
Platform
Hydrostatic skid-steer
Sensing
Fused LADAR + RADAR
Terrain
Unimproved, difficult

CMU National Robotics Engineering Center

Commercialization Specialist · 1999–03

I developed the path tracking and obstacle detection systems for a series of automated golf course fairway mowers — 2–3 cm precision to cut well-defined cross-hatch patterns, plus obstacle detection so the machines never hit anything unexpected on the fairway.