Autonomous Robotics
Researcher: Hugo Guterman
Department: Electrical & Computer Eng.
Faculty: Engineering Sciences
E-mail: hugo@ee.bgu.ac.il
The Laboratory for Autonomous Robotics (LAR) was founded in 1998. Due to the complexity and interdisciplinary nature of the issue, researchers and students of three Departments: Electrical and Computer Engineering, Mechanical Engineering and Industrial Engineering and Management work in close collaboration.
The main objective of LAR is to conduct basic and applied research resulting in technologies that support "state-of-the-art" capabilities for autonomous platforms. Towards this end LAR research concentrates on the areas of System Architecture, Sensors (Video, Sonar, Inertial, LADAR, GPS), Image Processing, Sensor Fusion, Vehicle Control, Path Planning and Obstacle Avoidance, System Integration, Wearable Computers, MMI, Communications, and Platform Design.
Researchers at LAR have successfully developed autonomous platforms for air, ground and sea, of different types and capabilities in the framework of several government-funded projects and in cooperation with Israeli industrial partners. Autonomous abilities including low-level control, path planning, navigation, obstacle detection and avoidance have been implemented on platforms such as an electrical wheelchair, a small tractor, on a TOMCAR dune-buggy vehicle and most recently on a Jeep-Wrangler. Tele-operation related issues developed in the laboratory include video transmission, advanced graphical interfaces and input (control) interfaces such as head-tracking and cyber-gloves.
In the field of cooperation among robots the research at LAR covered protocols for inter-vehicle communications and optimal mission planning under given constraints.
The most relevant international cooperation of LAR is with DARPA. Two projects are currently pursued within this cooperation: “Detection and avoidance of moving obstacles" and “Landmark-based navigation". LAR participated as a member of Team AvantGuardium in the DARPA Urban Grand Challenge. Team AvantGuardium was formed by US and Israeli defense R&D groups, who have been involved in the development and fielding of operational UGVs in real world applications.
Cognition, Perception and Learning Systems
Although, today's computers are cheaper, faster, easier to use and able to perform heavy calculation, their basic structures have not changed much during the last fifty years. The basic Von-Newman and Turing machine are nowadays faster, multitasked, vector oriented, etc., but they are unable to deal with basic cognitive tasks or complex non-linear systems.
Attempts to apply computer technology to image, speech, forecasting, control and fault detection share several characteristics: all require pattern recognition or the ability to identify, control or/and classify an entity or system despite noise, distortion and uncertainty. All hold great commercial promise because they reflect the human world. Finally, all are beginning to employ neural networks and fuzzy logic to improve accuracy, short the development cycle, and reduce cost.
Having been used experimentally for decades, neural networks and fuzzy logic are reputedly a solution in search of a problem. More recently, though, they have begun to be used in practical applications, and this trend can only accelerate now that more power computers, specialized hardware, ASIC, and software developments are available to speed product development.
The Laboratory of Fuzzy Logic and Neural Network Applications is involved in projects in the following areas:
- Image Processing Algorithms for ATR & ATD
- Intelligent Video Compression
- Speaker Diarization and Verification
- Algorithms and Hardware for Real Time Image Processing
- On the Fly Adaptive System
- Adaptive link assignment in a probabilistic environment
