In order to increase the terrain adaptability, the walking robot has to install visual sensor. The paper defines the visual sensor for walking robot as a MAp Realization System, abbreviated as MARS, and investigates the way to realize practical 3 D range finding device which forms the major part of the MARS.
Firstly the improvement of the range finder's performance to separate and extract projected laser slit light in ambient light is investigated. A video signal processing procedure named DDD (Dual signal extraction with Delay and Difference) method, which extracts the pulse signal by making use of the dual pattern of positive and negative pulse in differenced video signal, is proposed. The specific method of DDD, that extracts only the maximum pulse, is shown more effective and is called MaxDDD.
A range finder with the signal processing system having optical interference filter, MaxDDD video signal processor and noise reduction software applies for the signals of adjacent scanning lines is constructed and shown its high S/N ratio.
Secondly the structual and algorithmic consideration to realize real time compensation of the swinging motion of the walking robot and to generate the terrain map while walking is made. As a result it is shown that the position measurement of one point is executed within 240 micro second and thus enabled to generate a map in real time. The constructed MARS range finder device weighs 1. 8 kg and is designed compact.
By these considerations the paper shows the feasibility to realize a practical visual system for walking robot.
View full abstract