Multi-vehicle Experimental Platform for Distributed Coordination and Control

Faculty Contact: Jonathan How

Other Faculty: Col. Young

Students: Ellis King, Yoshiaki Kuwata, Chung Tin, Ian Garcia, and Luca Bertuccelli

Project funded by the DoD DURIP Program FY2002

Pictures From Recent Flight With 2 Autonomous UAVs

Movie of two Autonomous UAVs flying in formation (2.5 MB) or (5 MB)

Movies from the onboard camera (1) (2) (3)

This project will build a multi-vehicle testbed to demonstrate and evaluate the coordination and control approaches under development at MIT as part of several ongoing DoD funded research programs (e.g., SEC and MICA).

The testbed will be used to address a variety of technical issues:
  • Distributed command and control algorithms
  • Dynamically reconfigurable network topologies
  • Resource allocation using hierarchical networks
  • Vehicle/fleet autonomy
  • Multi-vehicle operations with humans-in-the-loop.
A unique feature of the proposed testbed is that it has been designed to reflect the complexity expected in future combat operations. In particular, the testbed includes many (semi-) autonomous vehicles that operate outdoors and have differing dynamics and capabilities. The vehicles will have onboard computers, communication capabilities, and various payloads. These features will enable an in-situ demonstration using realistic sensors (e.g., GPS), payloads (e.g., cameras), and vehicle separations (e.g., 1-1000m).

To reduce the operational complexity, the fleet will consist of several vehicles from four classes (UAV's, helicopters, blimps,rovers). This provides a good combination of flexibility, agility, and mobility, and should permit the use of the testbed in a broad range of applications. In particular, the fleet could be used to investigate coordinated action in restricted areas (urban or under a forest tree canopy). We have purposefully selected a broad array of vehicles so that many of them can be ``flown'' together in a fleet that can be simultaneously tasked to perform multiple missions.

Of particular significance is the goal of having many of these vehicles working in coordinated fashion to provide surveillance, detection, assessment, and tracking of multiple moving targets. This will then provide an excellent facility for demonstrating advanced coordination and control concepts in real-time, in a realistic environment.

Vehicles that have been developed so far:

  Automated UAVs
Recent Flight Data 1 UAV
and 2 UAVs

Using the Piccolo Autopilot
from CloudCap Technologies
Autonomous rovers (8)
and 4 blimps
Recent rover experimental data

Recent Work:

Demonstrated successful autonomous flight on the first attempt.
Small video clip (286K) from second flight. Total of ~16 min of autonomous flight.
Recent work has focused on hardware-in-the-loop simulations using three of the Cloudcap autopilots.
New setup
  Hardware-in-the-loop testbed. The 3 autopilots are stimulated by sensor data exactly as they would be in the air. They perform the waypoint control following a trajectory that is communicated to them from the ground station via the 900 MHz link. All control is currently done on the ground and uplinked.
Typical results showing a cooperative attack around the obstacles of 3 UAVs against the targets. The results (e.g. X,Y,Z,V,P,Q,R,) from each vehicle are then logged and movies of the scenario are made in AVDS.

Results from hardware in the loop tests on typical scenarios. These results demonstrate a
successful integration of the MILP-based planning algorithms with the UAV testbed

Last updated: April 1, 2004