Spike based Learning and Control for Multimodal Agile Sensory Integration and Behavior
Abstract
Spike based Learning and Control for Multimodal Agile Sensory Integration and Behavior Flying insects are able to perform impressive acrobatic feats such as simultaneously sensing and avoiding predators and feeding from moving flowers, all the while stabilizing flight subject to both physical and sensory disturbances, such as winds and poor luminosity. Although insect studies have inspired interesting and sometimes useful solutions to simplified robot control and coordination problems, the level of autonomy and adaptation observed in behaving insects remains unparalleled in artificial systems. Very few examples of fully autonomous flight systems, i.e., relying solely on onboard sensors and processing with no human in the loop, exist, and they typically involve slow maneuvers and simple objectives, such as following a predefined trajectory; or, agile maneuvers in highly controlled environments with accurate inertial sensory feedback. Despite clear evidence that animal sensorimotor control and navigation skills far exceed those of artificial systems, few animal studies have resulted into architectures that are applicable to robots. Previous animal studies on neural feedback control mechanisms have focused on obstacle avoidance and navigation in flying or walking insects. However, obstacle avoidance and navigation are among the best understood problems in robotics and autonomous systems, and fully autonomous flight in complex unstructured environments is known to require far more diverse perception and control capabilities.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 14, 2022
- Source ID
- FA95501910396
Entities
People
- Silvia Ferrari
Organizations
- Air Force Office of Scientific Research
- Cornell University
- United States Air Force