Mobile Robot Onboard Accelerated Computation

Abstract

Real~time robot visual and navigation learning requires extensive computational resources. Racks of general~purpose computer processors fill this role but relegate their robots to controlled environments due to the weight, size, and power consumption of server farms. FPGAs contain the ability to dynamically assemble into accelerators that can be optimized for specific algorithms. The massive parallelism and low power of FPGA circuitry provides the opportunity to mobilize highlyparallel algorithms, i.e. those with high Amdahl fractions, which is a generally uncommon trait for software but known to be inherent in brain~circuit algorithms. It is being proposed by the team to mobilize a set of well~studied brain~circuit algorithms by adapting them for execution on a novel low~power Processor~FPGA design; the resulting device will be tested on mobile robots for its ability toaccelerate learning and increase battery life in on~board robot use.

Document Details

Document Type
DoD Grant Award
Publication Date
Aug 12, 2016
Source ID
N000141612359

Entities

People

  • Richard Granger

Organizations

  • Board of Trustees of Dartmouth College
  • Office of Naval Research
  • United States Navy

Tags

Fields of Study

  • Computer science

Readers

  • Agent-Based Social Robotics and Mobile-Assisted Learning in Virtual Environments.
  • Integrated Circuit Design and Technology.
  • Parallel and Distributed Computing.

Technology Areas

  • AI & ML
  • AI & ML - Autonomous Systems
  • AI & ML - Neural Networks
  • Autonomy