Imminent Collision Detection

Passive Collision Detection (PCoDe) for Sense and Avoid

PCoDe predicts imminent collisions using only a passive monocular camera. This technology builds upon UC's unique passive ranging technology M2PARS to track other aircraft. Together with M2PARS, PCoDe can provide a passive-only end-to-end Sense and Avoid solution for UAS.

imminent_collision_detection

Technology description

  • PCoDe is based on probabilistic trajectory prediction along with estimation of the closest point of approach (CPA), the distance at CPA and the probability of collision
  • Extensive Monte Carlo simulation results of PCoDe have shown promising collision detection performance with high correct detection rate and low false alarm rate (above 95% correct detection rate with 2% false alarm rate)
  • High fidelity camera sensor noise models are used to improve performance
  • PCoDe can be easily adapted for unmanned surface vehicle use

UC has demonstrated the system (detection, tracking and ranging - M2PARS) offline using MAV test bed available at BYU. The following video demonstrates the results.

UC is currently working with DARPA and AFRL to develop and test an end-to-end SAA solution. We are currently conducting flight tests on a Group 3 UAS. The following video demonstrates our results from one of our flight tests conducted in January 2016.