Goran Huskić

Fast and safe autonomous outdoor navigation of mobile robots is an open and challenging problem, with various possible applications, such as agriculture, search and rescue, exploration or inspection. Furthermore, human-robot cooperation (such as person following) in rough terrain conditions, dynamic environments, and at higher speeds makes the task even more challenging.

For the purpose of this research, we use two skid-steered mobile robots Summit XL from the company Robotnik. Controlling a skid-steered vehicle is additionally challenging, since its motion cannot be described with pure rolling, without slipping and sliding.

Fig.1 One of our Summit XL robots, front and back view

We propose a full navigation system for outdoor person following at higher speeds. Our system includes path planning, path following control and obstacle avoidance. It allows the robot to follow a jogger in various outdoor scenarios, fully autonomously.

Demonstration of our new navigation system

Path following focuses on minimizing the distance and orientation error between the robot and a geometric reference path. Our results allow accurate path following at higher speeds, on different terrain types, both indoors and outdoors. A demonstration of the proposed controller on a Robotnik Summit XL robot can be seen in the video below.

Demonstration of our path following controller on a Robotnik Summit XL robot:

As a product of collaboration with LAAS-CNRS in Toulouse, France, the proposed controller is evaluated on a Segway RMP 440 robot as well. A video demonstrating these experiments is shown below.

Demonstration of our path following controller on a Segway RMP 440 robot

As seen in the videos, the controller is evaluated on different and challenging terrain types. The experiments were conducted at higher speeds: with Robotnik Summit XL at speeds up to 2.5 m/s, and with Segway RMP 440 at speeds up to 6.16 m/s. The proposed controller outperformes two other state-of-the-art-algorithms in all cases. Details can be found in [2]. Example results can be seen in Fig. 2 and Fig.3.

Fig.2:Results from one of the experiments on a grassy flat terrain with Robotnik Summit XL: red - desired path, blue - actual path; path length: 159.83 m, mean error = 6.9 cm, max error = 22.36 cm, mean speed = 2.15 m/s, max speed = 2.54 m/s

Fig.3:Results from one of the experiments on an uneven grassy terrain with Segway RMP 440: blue - desired path, red - actual path; path length: 400.14 m, mean error = 12 cm, max error = 44.19 cm, mean speed = 3.12 m/s, max speed = 3.54 m