Options
Moya-Albor, Ernesto
Vision-Based Autonomous Navigation with Evolutionary Learning
Bio-inspired Optical Flow-based Autonomous Obstacle Avoidance Control
2020
,
Moya-Albor, Ernesto
,
Ponce, Hiram
,
Brieva, Jorge
,
Coronel, Sandra L.
,
Chávez Domínguez, Rodrigo
In this paper, we propose a vision-based autonomous robotics navigation system, it uses a bio-inspired optical flow approach using the Hermite transform and a fuzzy logic controller, the input membership functions were tuned applying a distributed evolutionary learning based on social wound treatment inspired in the Megaponera analis ant. The proposed method was implemented in a virtual robotics system using the V-REP software and in communication con MATLAB. The results show that the optimization of the input fuzzy membership functions improves the navigation behavior against an empirical tuning of them. © 2020, Springer Nature Switzerland AG.
2019
,
Moya-Albor, Ernesto
,
Gomez-Coronel, Sandra L.
,
Ponce, Hiram
,
Brieva, Jorge
,
Chávez Domínguez, Rodrigo
,
Guadarrama-Muñoz, Alexis E.
In this paper, we propose a new methodology for autonomous obstacle avoidance control using a bio-inspired optical flow estimation. The main difference with other methods is that we use an image model inspired by the human vision system to define the constraints in the optical flow formulation which includes a Hermite transform (HT) and a perceptive mask. We use a physical robot platform to test the control algorithm, where due to the structure of the chassis a forward, reverse and turn movements were defined. The robot has a RBG camera to capture images of the path and then calculate optical flow estimation. To define velocity and direction robot response we propose a fuzzy controller. Finally, we made some experiments to demonstrate the performance of control navigation, and how responds algorithm using HT and a perceptive mask. © 2019 IEEE.