Coverage Efficiency in Autonomous Robots With Emphasis on Simultaneous Localization and Mapping Mo Lu Computer Systems Lab Q2
Abstract Coverage Efficiency is a major goal in autonomous systems Project approaches CE using SLAM Using SLAM, a autonomous system will be able to map and process an environment for efficiency
Introduction Today, automated systems have supplemented humans in previously labor-intensive tasks. Automated lawnmowers are an example of these systems, but the currently available technology in automated lawnmowing is inefficient and primitive. This project will propose and implement an alternate method to automated lawnmowing, known as Simultaneous Localization and Mapping, then report back the results.
Background Modern commercial autonomous lawnmowers (ALM's) are grossly inefficient in terms of runtime and coverage Random cuts and turns Dummy sensing Previous work in the field using SLAM include the annual Ohio University robotic lawnmower competition Problems of runtime v. coverage Military applications
SLAM Theory Scan for obstacles via laser scanner or similar device Update scans until entire map can be created, ie: all boundaries and obstacles connect Create obstacle and boundary map using scan outputs Analyze map via recursive run-through to determine most efficient path Run optimal path
Discussion: What's Been Done and What it Means Matrix-based environment simulation – Environment is pre-created, obstacles, boundaries and size have been set Robot keeps track of location Pings in 180 degree field of vision Returned data forms obstacle map Map is cross checked with environment for accuracy Results indicate that the scanning and mapping code works with various obstacles Further adaptations are needed before mapping works in live environments Need to address more realistic conditions – Power sources – Terrain – Complex polygonal navigation
Results Q2
Results Q2 Cont.
Program Running Screenshot
Other Obstacles
Conclusions and Plans Scan mimicking works, as does matrix mapping Adapt program for random matrices Incorporate more graphics Adapt program for terrain types (unmowable v. mowable grounds) Adapt program for use with LMS rangefinder -Python to C++