RICERCA ITALIANA     

PICTURE: Planning and Intelligent Control Techniques for Unstructured Robotic Environments

Università degli Studi di Cassino

Abstract

The goal of the research program is to coordinate the efforts of part of the national control systems community operating in the robotics domain toward the development of planning and control techniques for robotic systems required to execute the assigned task in unstructured environments. The issues considered play a crucial role in raising the degree of autonomy needed to accomplish a robotic task in all those cases, frequent in servicing applications, in which a-priori knowledge of the environment state is low (e.g., in presence of moving bodies other than the robot itself) or null (e.g., in exploration missions). In this scenario, the main topics that will be investigated in the research program are sensor fusion techniques, simultaneous localization and map building algorithms, multi-sensory feedback control techniques, locomotion systems for uneven terrain, motion coordination of vehicle-manipulator systems, and motion planning and control of robot teams composed of multiple autonomous vehicles. Special emphasis will be finally paid to the issues related to robotic applications in underwater environments.

Principal Investigator

Stefano CHIAVERINI Università degli Studi di CASSINO

Research Objectives

The development of robotic systems for service applications requires the solution of several problems among which those related to the autonomous and safe execution of the task stand out when a-priori knowledge of the environment is intrinsically low or null at all. This is, for instance, the case of tasks to be executed in cluttered environments or just consisting in the exploration of the environment itself. It is evident that, in this scenario, both planning and control of the robot must be performed in a fundamentally different manner from that suitable to much structured contexts, such as those typically characterizing conventional industrial applications.

The Research Program PICTURE is thus aimed at the development and the experimental validation of planning and control techniques for robotic systems executing tasks in unstructured environments. In this framework, the main lines characterizing the research activity subject of the Program are the following:

* sensor fusion techniques, to exploit at one time different sensory capabilities at their best;

* simultaneous localization and map building algorithms, to dynamically adapt the motion planning;

* multi-sensory feedback control techniques, to provide proper reaction ability to different environmental conditions;

* adoption of non-conventional locomotion systems, to allow mobility also on unstructured terrains typical in outdoor >>>

First Results

Setting up of the site http://webuser.unicas.it/picture ensuring centralized access to the Program information and results, besides to the resources made available by the single Units.

State-of-art update.

Definition of the results achievable in view of the obtained resources.

Specification of the updating needs of the experimental setups.

Coordination meeting.Development of new methodologies.

Prototyping of devices and software where expected.

Preliminary implementations.

Purchase of the requested equipments.

Scientific publications submitted to international conferences.

First Program meeting.Experimental and/or simulation testing of the proposed solutions.

Further methodological developments and progress in the prototyping activity.

Scientific publications presented at international conferences and submitted to international journals.

Coordination meeting.

Definition of the structure of the Program monograph and starting of the editorial agreements concerning its publication.Final experimental testing and validation of the proposed solutions.

Conclusion of the prototyping activity.

Publications in scientific journals and proceedings of international conferences.

Final implementation of the Program web site, allowing access to >>>

Timescale

24 months

National and international background

The adoption of robotic systems in unstructured environments requires the solution of several problems now at the border of the scientific research in Robotics. The autonomous execution of a task in a scenario where the environmental context is unknown beforehand or is subject to continuous and hardly predictable variations, raises a difficult challenge which asks to trade off operation efficiency with prevailing safety reasons, both of the robotic agent itself and of the other bodies sharing the environment with it, and quality of the execution; of course, these must not compromise the mission accomplishment. The complex nature of the problems involved in the use of a robotic system requires synergy of a broad spectrum of scientific expertise, with consequent broad spectrum of concurrent methodologies and technologies. With special reference to the motion planning and control aspects that are the subject of this Research Program, in the following we restrict ourselves to outline the scientific background of the seven Topics defined at point 2.1, highlighting the most significant references. Further detail can be found in Model B of the single Units.


FUSE topic: Sensory data fusion

Sensor fusion techniques are widely used to combine information coming from several sensors (of the same or different kinds) and, possibly, other source of knowledge, with the aim of improving the accuracy of measures or inferring more specific results than >>>