Event-based Integrated Sensorimotor Planning and Control for Insect-scale Robots

Abstract

This project seeks to develop event-based control and navigation methods for autonomousinsect-scale robots that utilize proprioceptive and exteroceptive sensory feedback to interactwith complex environments. Physics-based mathematical models of the insect-scale robots andtheir environmental conditions will be used to develop a high-fidelity 3D simulation that willbe used to first develop" and, later, refine the event-based sensorimotor controllers viaperturbative learning. These new sensorimotor architectures will be" developed for andtested on custom insect-scale robots equipped with a host of onboard sensors including eventbasedvision sensors" inspired by insect ocelli, low-power gyroscopes, optical flow sensorscomprised of elementary motion detectors (EMDs), force sensor""s with functions similar toinsect campaniform sensilla, vision sensors, and attitude and wind velocity sensors inspired byinsect a"ntennae. The robots and sensors will be fabricated by the Wood laboratory fortesting in physical experiments that will parallel the" studies performed in simulation.Because event-based controllers can be designed without the need for an accurate robotmodel, meth""ods such as perturbative learning will be used to adapt the sensorimotor controllersto complex disturbances, maneuvers, and sensing" missions without requiring extensive modelingor system identification efforts. By reducing the need for parameter/system identific"ation, it isalso expected that less time and costs will be required for robot design and manufacturing whichinevitably introduce p"arameter variations from robot to robot. The expected outcome of thisresearch project are terrestrial robots able to detect substra"te disturbances by leg proprioceptorssuch as the campaniform sensilla, and aerial robots that can use antenna-like sensors to detec""ttransient wind gusts and alter motor patterns locally. Large disturbances such as obstacles,sustained substrate movements, or com""plex flows due to wall or ground effects will be measuredby onboard low-power gyroscopes, elementary motion detectors, and attitude"" and wind velocitysensors, and combined with visual cues detected by ocelli for path integration and re-planning.Furthermore, low-" to high-resolution millimeter-scale cameras will be utilized inconjunction with event-based sensorimotor controllers to develop te"rrestrial and aerialinsect-scale robots capable of tracking fixed or moving targets, while simultaneouslynegotiating obstacles and" disturbances that may require fast and agile maneuvers.

Document Details

Document Type
DoD Grant Award
Publication Date
Jun 09, 2017
Source ID
N000141712614

Entities

People

  • Silvia Ferrari

Organizations

  • Cornell University
  • Office of Naval Research
  • United States Navy

Tags

Readers

  • Data Mining and Knowledge Discovery.
  • Distributed Systems and Data Platform Development
  • Robotics and Automation.

Technology Areas

  • AI & ML
  • AI & ML - Autonomous Systems
  • Autonomy
  • Autonomy - Autonomous System Control