I. Introduction
The automation of robots and vehicles is becoming increasingly important in many different industries. Companies are trying to retrofit existing vehicles and develop autonomous assistance functions or complete autonomy skills. Most vehicles and robotic systems are equipped with different sensor modalities to achieve the best possible sensor coverage and redundancy. However, most robotic systems lack the feedback of steadiness and stability. Legged robots are the exception, as pressure sensors are usually installed at the feet. The advanced and new generation of humanoid and legged robots has not only the versatility, but also the reliability and robustness. These features are achieved through some kind of force or torque control - either through integrated load cells in the joints or through pressure sensors on the end-effectors to properly control the interaction forces with the environment [1], [2]. Many applications which are to be automated can be addressed with ordinary sensors. For example, an inertial measurement sensor can be used to determine the position and orientation of the robot. When moving in rough terrain or climbing steep stairs > 45°, a feedback of the stability of the robot can be essential. Tracked vehicles, along with wheel-driven propulsion systems, are the preferred platform for UGVs in poor terrain conditions or harsh environments. However, a tracked vehicle concept also has disadvantages, especially during cornering maneuvers or turning maneuvers around the instantaneous center of rotation (ICR). Depending on the coefficient of friction between the tracks and the ground and the total weight and center of gravity of the vehicle, such models produce high slip and shift of the trajectory. A precise prediction of the resulting track on the basis of the current terrain situation would increase the control/steering stability of autonomous or even remote operated vehicles. Therefore this paper presents a tactile surface sensor to provide improved odometry determination for different ground conditions. The key question to be answered is how to determine the resulting position after actuating the right and left tracks independently to perform turns etc. or in other words: If a robot starts from a position , and the right and left tracks move respective distances and , the question is always: What is the resulting new position ? Odometry is obtained by integrating wheel encoders for each track which occurs following odometry error sources:
Limited encoder resolution (deterministic)
Belt coating (deterministic)
Unequal track diameter (deterministic)
Variation in the contact point of tracks (deterministic)
Unequal floor contact and variable friction can lead to slipping (non deterministic)
URDF model and stability vector (top); pressure distribution on bars (bottom)