We have improved autonomous navigation by a visual system and obstacle detection

November 26, 2019  |  Development

The Androver II robot, completely developed by RoboTech Vision, is able to move autonomously in an unfamiliar environment thanks to our algorithm. Unlike navigation with a local environment map, which we test on the Husky A200 robot, Androver II does not have to have an environment mapped in advance. It is controlled mainly by what it detects with its sensor system.

Controlled mainly by sensors

RoboTech Vision strives to develop the most robust solution that would not only work in any environment, but also be safe and deployable on as many devices as possible. This is also one of the reasons why the company is testing different types of autonomous navigation on different robotic platforms. After they managed to autonomously send the Husky A200 robot by Clearpath Robotics to the destination based on a map, they also began to improve the reactive navigation on the Androver II robot, which is partly used by Husky A200 too.

With its visual system, Androver II segments the road and intersections, recognizes obstacles and moves along the route without collisions. The destination for it is the point entered in the Open Street Map data, while the environment is not pre-mapped. This type of navigation is called reactive navigation because the robot responds to conditions in the environment. On the way to its destination, Androver II managed to segment the path despite leaves, while avoiding obstacles.