potential field based navigation for planetary rovers using internal states
- Paper number
IAC-08.C1.1.4
- Author
Mr. Mohamed Abdelwahid, University of Strathclyde, United Kingdom
- Coauthor
Prof. Colin R. McInnes, University of Strathclyde, United Kingdom
- Year
2008
- Abstract
1 Introduction and Problem Definition: In recent years new assumptions about architectures needed for intelligence have emerged. These approaches attempt to emulate natural, rather than artificial intelligence and are based on, or at least inspired by, biology. The local minima problem has been a serious issue for artificial potential field methods for robot path planning [1]. The problem of local minima has been discussed by many researchers. Our motivation is that most of these attempts are not suitable for real time applications [2]. The problem can be defined such that an artificial potential field at a goal position induces the rovers’ motion towards the goal. In order to prevent collision with a static obstacle, an additional repulsive potential field is required. In general, a local minimum may form due to the superposition of the goal potential and that of the obstacles, resulting in a group of rovers, becoming partially or globally trapped in a state other than the goal. 2. Internal State Model: Inspired from escape from complex workspaces, which can be seen in many natural systems, the use of dynamic internal states (potential field free parameters) is considered as a means of allowing agents to manipulate the potential field in which they are manoeuvring to solve the local minimum problem. 3. Model Analysis: We compare our model performance with the most updated technique to solve the local minimum problem; the forward chaining technique, by comparing the results obtained by the internal state model to the results obtained by forward chaining, published in [3], for the same environment. In addition to single rover, we also consider cooperation amongst multiple rovers. It is shown that the swarm aggregation concept is used and the rovers prefer to aggregate when facing a navigation problem in a way that matches the studies based on real animal group behaviour. 4. Summary of Results: The simulations results show that using the internal state model, a swarm of planetary rovers, rather than moving in a static potential field, are able to manipulate the potential according to their estimation of whether they are moving towards or away from the goal, which allows them to escape from and maneuver around a local minimum in the potential field to reach a goal. An application of a swarm of rovers to solve the problem for different shaped obstacles during a mission is introduced to show the efficiency of the model. 5. Conclusions: A new model for a swarm of rovers interacting via pair-wise attractive and repulsive potentials is presented. The model updates the state of the art in overcoming the local minima problem through solving the problem with comparatively lower computation cost than other methods. The model proves stable convergence to a goal and provides similarities with the behaviour of real groups of animals. References [1] Khatib, O. (1985). Real-Time Obstacle Avoidance for Manipulators and Mobile Robots. IEEE International Conference, Vol. 2, pp.500 – 505. [2] Abdelwahid, M. and McInnes, C. (2007). Solving the potential field local minimum problem using internal agent states. Submitted to the Journal of Robotics and Autonomous systems. [3] Bell G. and Weir M. (2004). Forward chaining for robot and agent navigation using potential fields. In Proceedings of Twenty-seventh Australasian Computer Science Conference (ACSC2004), pp. 265 – 274.- Abstract document
- Manuscript document
IAC-08.C1.1.4.pdf (🔒 authorized access only).
To get the manuscript, please contact IAF Secretariat.