Paolo Pirjanian

Multi-robot target acquisition using multiple objective behavior coordination

We propose an approach to multi-robot coordination in the context of cooperative target acquisition. The approach is based on multiple objective behavior coordination extended to multiple robots. It provides mechanisms for distributed command fusion across a group of robots to pursue multiple goals of multiple robots in parallel. The mechanisms enable each robot to select actions that not only benefit itself but also benefit the group as a whole. Experimental results with two mobile robots validate that, by using this method, a group of robots can successfully truck and acquire a moving target.

Multiple objective behavior-based control

Abstract The notion of optimality and its feasibility are revisited in the context of behavior-based control. It is argued that optimal behavior is not feasible for real-world applications. As an alternative to optimality I promote Pareto-optimal and satisficing solutions which correspond to efficient and “good enough” behavior. It is then demonstrated that multiple objective decision theory provides a suitable framework for formulating behavior-based controllers that generate Pareto-optimal and satisficing behavior. A set of simulated and real-world experiments are reported that support this view.

A decision-theoretic approach to fuzzy behavior coordination

We propose a novel approach to behavior-based control, which combines fuzzy logic with multiobjective decision theory. Fuzzy logic provides powerful tools for formulating sophisticated behavior-based systems for robot control. However, it suffers from problems associated with resolving behavioral conflicts, which are mainly due to the deficiencies of fuzzy inference techniques. Hence, we propose to formulate the inference mechanism using a multiobjective decision-theoretic approach, which provides formal tools for behavior coordination and resolving conflicts in a principled manner. Preliminary experimental results are reported.

Action selection within the context of a robotic colony

Preparation of planetary surface sites prior to a manned mission can be accomplished through the use of a robotic colony. The task of such a colony would include habitat deployment, setup of in-situ fuel and oxygen production plants, and beaconed road placement. The colony will have to posses a great deal of autonomy for this ambitious list. BISMARC is a behavior based system for the control of multiple rovers on planetary surfaces. During the past few years the system has performed well in multiple cache retrieval simulations, and a certain degree of fault tolerance has been included in the design. In this paper we address the extensions to BISMARC that would be necessary for a robotic colony application. These extensions include a wider array of behaviors, better communication and mapping capabilities, and fault tolerance shared by the colony. The results of some simulations for habitat site preparation are reported.

Ant-inspired navigation in unknown environments

In contrast to most other ant species, desert ants (Cataglyphis fortis) do not use pheromones to mark their path. When returning from a foraging trip to their nest they navigate both by path integration and by visual landmarks. An egocentric navigation system based on path integration alone su ers from two major pitfalls: 1. It must run uninterrupted as long as the animal moves and 2. It is inherently susceptible to cumulative navigation errors. Hence, if the animal visited an area repeatedly, it is advantageous to take an occasional positional x by acquiring landmark-based geocentered (alocentric) information. Indeed desert ants and honeybees use such information in addition to path integration. On a familiar route, when ants can steer by visual landmarks, they adopt a xed and often circuitous path consisting of several separate segments that point in di erent directions. Such multi-segment journeys are composed partly of stored local movement vectors, which are associated with landmarks and are recalled at the appropriate place. During the experiments reported in [4] and [1], ants collected right before entering their nest after a foraging trip are deprived of their global vector. When these ants were placed at points along their familiar route, they were able to use previously seen visual features in order to return to their nest. It is believed [1], that two dimensional visual snapshots of landmarks along the homing path are stored in the memory of the ants and upon recognition they trigger a local vector that describes the transition to the next landmark on the way to the nest. These local vectors are also calculated based on odometry. They are links that point towards the direction of the next landmark along the path and are proportional to the distance between consecutive locations. Due to their limited length compared to the global vector the local transitions are more precisely learned and reproduced during foraging trips. The vision-based (local) navigation in ants works in parallel with global navigation. In order to ensure that an alternative navigation mechanism is available, desert ants use landmark-based navigation as the main method along familiar paths. They resort to global navigation only if the former fails, that is if in a dynamic environment landmarks are missing or are being occluded by new ones. If one of