Tuesday, 22 September 2009

Journal Paper: Engineless Unmanned Aerial Vehicle Propulsion by Dynamic Soaring

Deittert, M; Richards, A. G.; Toomer, C. A. & Pipe, A., Engineless Unmanned Aerial Vehicle Propulsion by Dynamic Soaring, AIAA Journal of Guidance Control and Dynamics, Vol. 32, No. 5, September–October 2009, p.1446.

Abstract: Dynamic soaring is a flight technique which extracts energy from wind gradients with the potential to power small unmanned aerial vehicles in maritime applications. Wind gradients of the required magnitude naturally occur at the air–sea interface due to friction between the waves and the moving air. Suitability of dynamic soaring as a means of propulsion requires clarification of the achievable flight performance and the likelihood of favorable winds. Optimal trajectories for minimal and maximal wind conditions are generated as well as trajectories for optimal cross-country travel. The flight model’s differential flatness property is used to simplify the optimization problem. The likelihood of favorable winds is predicted based on long term weather statistics and knowledge of the minimal and maximal permissible wind strengths. Comparison of the likelihood of favorable winds for the wandering albatross and an unmanned aerial vehicle of similar size shows that the ability to fly close to the surface is a key factor governing dynamic soaring performance.

Journal Paper: Path-Planning with Avoidance Using Nonlinear Branch-and-Bound Optimization

Eele, A.J. & Richards, A.G., Path-Planning with Avoidance Using Nonlinear Branch-and-Bound Optimization , AIAA Journal of Guidance, Control and Dynamics , Vol. 32 , No. 2 , March 2009 , Pages 384--394 .

Abstract: This paper describes a novel method for finding optimal trajectories for a vehicle constrained to avoid fixed obstacles. The key property of the method is that it provides globally optimal solutions while retaining the full nonlinear dynamics model. Applications for the method include guidance of unmanned aerial vehicles, air traffic control and robot path planning. The core concept is the direct application of branch-and-bound optimization to find guaranteed, globally optimal solutions to nonconvex problems. The method tailors the branch-and-bound approach specifically for avoidance problems by exploiting two new ideas: first, using a geometric branching strategy based on the decision between passing an obstacle clockwise or counterclockwise; and second, solving the resulting optimization. The algorithm is refined by comparing nine geometric branching strategies. The solution time of the method depends on the choice of branching strategy, which determines how the solution tree is explored. A good branching strategy is one requiring fewer tree branches to be enumerated before the global optimal is found. The best of these branching strategies has been compared with an existing mixed-integer linear programming approach and demonstrated a significant improvement on mixed-integer linear programming solve times.

Dancing robots

Thanks to the efforts of various project students over the years, distributed predictive control finally works on the robot trucks. What you see here is completely autonomous: we just stand at watch. Using its on-board computer and MILP optimization software, each truck plans its next 15 seconds of movement every 3 seconds, and then transmits its plan to the others. They know their own locations from an ultrasound positioning system, and communicate by wireless network. The trucks plan out of phase, to ensure there's time for the messages to be passed around. The "tube" approach to robustness helps them withstand uncertainty.

I've moved

Thanks to building works and equipment updates, I have a new office and phone number:

Email: arthur.richards@bristol.ac.uk

Tel.: +44 (0) 117 331 5534 (internal: 15534)
Fax: +44 (0) 117 927 2771

Room 0.54A
Department of Aerospace Engineering
University of Bristol
Queens Building
University Walk

Blog Renaissance

I'm renewing the experiment of running my research website in blog form, rather than the more traditional web page format. The plan is to cover all the usual material - publications, group members, etc - plus some more news-like material, such as breakthroughs, demonstrations, videos, etc., and all in a convenient format. Fingers crossed...