Related Experiment Videos
Sensory architectures for biologically inspired autonomous robotics.
1Department of Electrical and Computer Engineering, The University of Arizona, Tucson 85721, USA. higgins@ece.arizona.edu
The Biological Bulletin
|May 9, 2001
Summary
Engineers can benefit from studying biological systems. Custom neuromorphic hardware enables efficient, low-power sensory systems for robots by emulating biological computational strategies.
Area of Science:
- Engineering
- Neuroscience
- Robotics
Background:
- Biological neural systems exhibit complex, real-time sensory and motor capabilities surpassing artificial systems.
- Common design strategies exist across diverse biological systems, offering valuable engineering insights.
- Conventional processors struggle to implement continuous-time, parallel biological architectures efficiently.
Purpose of the Study:
- To explore the benefits of studying biological systems for engineers.
- To highlight the limitations of conventional processors for biologically-inspired architectures.
- To introduce custom neuromorphic hardware as a solution for efficient implementation.
Main Methods:
- Investigating common design strategies in biological neural systems.
- Developing custom neuromorphic hardware to emulate biological architectures.
- Implementing neuromorphic hardware for sensory systems in autonomous robots.
Main Results:
- Neuromorphic hardware provides power- and space-efficient implementation of biological computational strategies.
- Demonstrated a low-level neuromorphic emulation of a visual motion detector.
- Developed a large-scale system-level spatial motion integration system using neuromorphic hardware.
Conclusions:
- Biologically-inspired computational architectures require specialized hardware like neuromorphic systems.
- Custom neuromorphic hardware is suitable for low-power, compact sensory systems in autonomous robots.
- Neuromorphic hardware enables the realization of complex sensory processing inspired by neurobiology.