SBIR Phase I: Aggressive Maneuvering of Small Autonomous Robots in Home Environments

Petronics Inc, Champaign IL

Investigators

Abstract

The broader impact/commercial potential of this project is to enable mobile robots to coexist harmoniously with people in their homes and offices. The market for consumer and office robots is projected to grow 17% annually, seven times faster than the market for manufacturing robots, reaching $1.5B by 2019. An important step toward market growth is creating autonomous robots that are unobtrusive, intelligent, and highly agile. Accomplishing this requires robots to be small enough to stay out of the way and fast enough to elegantly avoid humans and environmental obstacles. Inappropriate noise levels and safety concerns make it unlikely that airborne vehicles will be prevalent in indoor environments, whereas wheeled mobile robots can achieve near-silent operation. Tiny, fast robots are unobtrusive enough to use as a low-cost surveillance tool in home or office security, and portable enough for covertly investigating hostile situations. People with severe disabilities could travel vicariously by combining a virtual reality headset with a telepresence robot. Fast maneuvering robots could be used as a compelling educational or entertainment platform for kids and adults. This Small Business Innovation Research (SBIR) Phase I project aims to prove the feasibility of enabling small wheeled robots to maneuver aggressively and predictably in varied operating conditions using consumer-affordable hardware components. While this project develops algorithms using a low-cost camera solution for localization, the methodology developed in Phase I will only improve performance as the technology for localization and navigation matures. The key challenges in creating a robot that can quickly navigate varied environments, as demonstrated during rigorous testing of early prototypes in real homes, involve understanding how a small robot moves on varied surfaces in the presence of slip, and correspondingly, how to accurately and efficiently plan predictable maneuvers on these surfaces. Three key Phase I objectives will address these challenges: 1) automatically learning surface models in unknown environments, 2) planning and executing aggressive maneuvers on learned surfaces, and 3) integrating a robot and a low-cost computer vision based localization system for autonomous control.