Reactive and continuous space-time control of multi-drone-fleets

System for continuously controlling quadrotor drone behavior using Signal Temporal Logic (STL) as the mission specification language

Problem:

Autonomous drones are increasingly being used for advanced applications such as package delivery, surveillance, and infrastructure monitoring. However, the requirement of complex instructions for spatial, temporal, and reactive navigation makes controlling drone systems challenging. Most existing encoding systems are simplified and limited, and do not consider timing constraints, scalability, or real-world behaviors.

Solution:

Dr. Mangharam and colleagues at the University of Pennsylvania have designed a new algorithm to control quadrotor drone behavior continuously using Signal Temporal Logic (STL). The method enables the consideration of explicit timing requirements and use of the full expressiveness of STL to optimize a smooth trajectory for drone movement. The system offers continuous-time STL satisfaction, feasible movement trajectories, and real-time drone control with scalability, providing a control framework for mission planning and execution for fleets of quadrotors.

Multiple autonomous drone missions in an urban environment. (Figure adapted)

Advantages:

Avoids oversimplifying abstractions while using off-the-shelf sensors and limited computation
Handles complex spatial, temporal and reactive requirements for drone missions
Scales well for problems involving up to 16 drones
Maximizes robustness in generating trajectories for quadrotors in continuous-time

Stage of Development:

Proof of concept on 16 drones in simulation and with 8 actual quadrotor drones.

Reference Media:

Pant, YV et al; 2018 ACM/IEEE 9th International Conference on Cyber-Physical Systems (ICCPS), 2018: 186
Intellectual Property:
Patent Pending

Desired Partnerships: