Bottom-up cognitive analysis of bionic inspection robot for construction site

Thomas Bock, Alexey Bulgakov, Sergei Emelianov, Daher Sayfeddine

Research output: Contribution to conferencePaperpeer-review

Abstract

Artificial intelligence aims to make robots more adaptive and versatile depending on the surrounding operating atmosphere. It permits the autopilot of the robot to generate optimal trajectory with reference to the energy efficiency criteria and risk avoidance in order to lead to the end-effector or the working element of the robot to the desired position, thus to perform safely the assigned task. One of the problems faced during the optimization of algorithms of the central autopilot is the overlearning, similar training sets, permanent operating conditions, which may lead to controversial result: non-adaptability of the robot. This can be clearly seen when training nonlinear model of neural network with exogenous inputs aiming to resolve the extrapolation of the movement function of a mobile agent (nontraditional desired trajectory). Hence the quick change in the operating conditions can lead to undesired outcomes. In this paper we analyze robot control performance based on situational approach, described using discrete mathematics operators and state-base/history base, time-base/action-base functions. The knowledge representation will be used to train auto-regressive neural network using situational time series.

Original languageEnglish
Pages74-78
Number of pages5
StatePublished - 2017
Externally publishedYes
Event34th International Symposium on Automation and Robotics in Construction, ISARC 2017 - Taipei, Taiwan, Province of China
Duration: 28 Jun 20171 Jul 2017

Conference

Conference34th International Symposium on Automation and Robotics in Construction, ISARC 2017
Country/TerritoryTaiwan, Province of China
CityTaipei
Period28/06/171/07/17

Keywords

  • Artificial intelligence
  • Cognitive robot
  • Narx neural network
  • Regression

Fingerprint

Dive into the research topics of 'Bottom-up cognitive analysis of bionic inspection robot for construction site'. Together they form a unique fingerprint.

Cite this