Planning the path of an autonomous, agile vehicle in a dynamic environment is a very complex problem, especially when the vehicle is required to use its full maneuvering capabilities. Recent efforts aimed at using randomized algorithms for planning the path of kinematic and dynamic vehicles have demonstrated considerable potential for implementation on future autonomous platforms. This paper builds upon these efforts by proposing a randomized path planning architecture for dynamical systems in the presence of fixed and moving obstacles. This architecture addresses the dynamic constraints on the vehicle's motion, and it provides at the same time a consistent decoupling between low-level control and motion planning. The path planning algorithm retains the convergence properties of its kinematic counterparts. System safety is also addressed in the face of finite computation times by analyzing the behavior of the algorithm when the available onboard computation resources are limited, and the planning must be performed in real time. The proposed algorithm can be applied to vehicles whose dynamics are described either by ordinary differential equations or by higher-level, hybrid representations. Simulation examples involving a ground robot and a small autonomous helicopter are presented and discussed.

/pdf/real-time-motion-planning-for-agile-autonomous-vehicles-4l6obiq3kn.pdf

Real-Time Motion Planning for Agile Autonomous Vehicles

Planning the path of an autonomous, agile vehicle in a dynamic environment is a very complex problem, especially when the vehicle is required to use its full maneuvering capabilities. Recent efforts aimed at using randomized algorithms for planning the path of kinematic and dynamic vehicles have demonstrated considerable potential for implementation on future autonomous platforms. This paper builds upon these efforts by proposing a randomized motion planning architecture for dynamical systems in the presence of fixed and moving obstacles. This architecture addresses the dynamic constraints on the vehicle's motion, and it provides at the same time a consistent decoupling between low-level control and motion planning. Simulation examples involving a ground robot and a small autonomous helicopter, are presented and discussed.

/pdf/real-time-motion-planning-for-agile-autonomous-vehicles-4eeud9gl71.pdf

Real-time motion planning for agile autonomous vehicles

The operation of an autonomous vehicle in an unknown, dynamic environment is a very complex problem, especially when the vehicle is required to use its full maneuvering capabilities, and to react in real time to changes in the operational environment. A possible approach to reduce the computational complexity of the motion planning problem for a nonlinear, high dimensional system, is based on a quantization of the system dynamics, leading to a control architecture based on a hybrid automaton, the states of which represent feasible trajectory primitives for the vehicle. The paper focuses on the feasibility of this approach, in the presence of disturbances and uncertainties in the plant and/or in the environment: the structure of a robust hybrid automaton is defined and its properties are analyzed. In particular, we address the issues of well-posedness, consistency and reachability. For the case of autonomous vehicles, we provide sufficient conditions to guarantee reachability of the automaton.

/pdf/robust-hybrid-control-for-autonomous-vehicle-motion-planning-3jwg5yiuul.pdf

Robust hybrid control for autonomous vehicle motion planning

Unmanned aircraft systems have the potential to perform many of the dangerous missions currently flown by manned aircraft, yet the complexity of some tasks, such as air combat, have precluded unmanned aircraft systems from successfully carrying out these missions autonomously. This paper presents a formulation of a level-flight fixed-velocity one-on-one air-combat maneuvering problem and an approximate dynamic programming approach for computing an efficient approximation of the optimal policy. In the version of the problem formulation considered, the aircraft learning the optimal policy is given a slight performance advantage. This approximate dynamic programming approach provides a fast response to a rapidly changing tactical situation, long planning horizons, and good performance, without explicit coding of air-combat tactics. The method's success is due to extensive feature development, reward shaping, and trajectory sampling. An accompanying fast and effective rollout-based policy extraction method is used to accomplish online implementation. Simulation results are provided that demonstrate the robustness of the method against an opponent, beginning from both offensive and defensive situations. Flight results are also presented using unmanned aircraft systems flown at the Massachusetts Institute of Technology's real-time indoor autonomous vehicle test environment.

/pdf/air-combat-strategy-using-approximate-dynamic-programming-19227wkpss.pdf

Air-Combat Strategy Using Approximate Dynamic Programming

Close formation flight is considered in the context of cooperative game theory and the natural aggregate motion of flocking birds, schooling fish, and herds of land animals. The airplanes in the formation are allowed to maneuver individually. However, the maneuver chosen by each airplane stems from basic flocking instincts such as collision avoidance, obstacle avoidance, and formation keeping. Consequently, there is no designated leader of the formation and the airplanes cooperate to achieve a common target seeking objective for the formation. Two examples are simulated. One example illustrates a formation of five aircraft completing a heading change maneuver while encountering several environmental obstacles. The second example demonstrates cooperative behavior as part of the formation breaks away to inspect a potential target or adversary. The results illustrate that the formation of aircraft can follow very precise and intricate three-dimensional paths while still acting as a group.

Formation flight as a cooperative game

A game-matrix approach has been developed to generate intelligent maneuvering decisions for low-flying aircraft during one-on-one air combat over hilly terrain. The decisions are made by comparing scores based on the predicted orientation, range, velocity, and terrain clearance for various maneuver combinations of both aircraft. A computer program implements the decision process in real time and computes the state of the aircraft employing the selected maneuvers. Predicted states are obtained by numerically integrating the dynamic equations. The program has been installed and demonstrated at the NASA Ames vertical motion simulator to provide an automated adversary for manned helicopter combat simulations. A stand-alone version of the program, in which the decisions of both combatants are generated by the computer, has also been prepared for nonpiloted simulations. Sample trajectories show that this new automated maneuvering logic generates some of the tactics employed by experienced pilots in combat flight tests. Whereas the current application applies to helicopter combat, the program can be used for fixed-wing aircraft by replacing those subroutines that provide the aircraft capabilities.

Game theory for automated maneuvering during air-to-air combat

A game-matrix approach is used to generate intelligent maneuvering decisions for nap-of-the-earth one-on-one helicopter air combat over hilly terrain. Decisions are made in real time using the AUTOMAN program by comparing scores based upon the predicted orientation, range, velocity, and terrain clearance for various maneuver combinations of both aircraft. The program has been demonstrated at the NASA Ames Vertical Motion Simulator, providing manned helicopter simulations. Sample trajectories from a stand-alone version of the program designed for nonpiloted simulations demonstrate that the present automated maneuvering logic generates some of the tactics employed by experienced pilots in flight tests. The program also has application to fixed-wing aircraft simulations.

Michael Falco

Papers

Game theory for automated maneuvering during air-to-air combat

Automated maneuvering decisions for air-to-air combat