Open search
Advertisement

Over hill, over dale: The six-legged robot does wander everywhere

Over rough and smooth: Lauron, the six-legged robot, developed by researchers at the FZI (Germany), has been granted with the University-Project-Award by Faulhaber, a German manufacturer of small electrical motors and motion controllers. Since many years, the robot has been improved. The Phasmatodea (also known as Phasmida) was the archetype. This stick insect or walking stick provides six legs, as the Lauron robot is doing. The current robot version, Lauron IVc, uses an embedded CAN network to link for example the motion control units to a Linux-PC.

OVER ROUGH AND SMOOTH: Lauron, the six-legged robot, developed by researchers at the FZI (Germany), has been granted with the University-Project-Award by Faulhaber, a German manufacturer of small electrical motors and motion controllers. Since many years, the robot has been improved. The Phasmatodea (also known as Phasmida) was the archetype. This stick insect or walking stick provides six legs, as the Lauron robot is doing. The current robot version, Lauron IVc, uses an embedded CAN network to link for example the motion control units to a Linux-PC.

The FZI engineers have developed their own motion controller boards. There are seven of these “Ucoms” modules installed in the robot. They are connected via the CAN network to the host-controller, a Linux-PC. The data transmitted by the PC includes the commanded positions of the three joint angles and the start/stop/reset command. The motion controllers report via the CAN network the actual motor encoder values, the 3D-force values of the feet sensors, the force value of the spring as well as the current in the joints. All these measurements are given as 3D-values. The used CXR 2657 series of graphite commutated DC motors by Faulhaber are controlled by means of the “Ucoms” modules. Each of the 20 installed small motors with a 26-mm diameter and a 57-mm length provide a constant torque of 35 mNm. Using gears and draw-wires, the total torque sums 20 Nm and peak torque values of up to 40 Nm. The motion-control units implement a speed-position mode. The behavior-based control, which controls the entire motion of the robot, is executed on the PC-104 host controller board. The image processing and environment modeling are performed on an additional PC-104 board. All software parts on the host controller, including the behavior-based control, are implemented in the modular MCA2 software.

The smallest part of the behavior-based control is the basic behavior modules. They are combined to achieve more complex motion behavior. In order to achieve a maximum of autonomy, each leg has its own group of four local leg behaviors. These behaviors create the needed swing and stance trajectory for each leg independently. The swing behavior creates the swing trajectory from the anterior extreme position (AEP) to the posterior extreme position (PEP). The stance behavior is responsible for the part from the PEP back to the AEP. The ground contact and collision behavior adjust the basic trajectory, when the ground is lost or a collision has occurred.

The local leg behavior groups are coordinated by different walking pattern behaviors (tripod, tetrapod, pentapod, and free gait). Depending on a given walking velocity vector and rotation value, a steering behavior creates the values for the AEP and the PEP for each leg. Then these two values are transferred to the local swing and stance behaviors. Three independent behaviors control the inclination, height and position of the main robot body with the purpose to ensure stability of the robot.

Walking in different terrain types requires appropriate walking parameters. For example, if the obstacles have become higher it is necessary to increase the swing height of the legs. With no environmental model the robot has no possibility to adjust its parameter to the terrain in advance. However, the behavior-based control provides many activity and reflection values, which deliver indirect information about the environment. Together with special sensors, these activity values are sued to create independent stability status information. For example, the sum of the activities of all posture behaviors is used to define one stability status value. A high sum activity results in a high status. A critical overall stability status stops the walking process and brings the robot in a stable position. This entire process is supported by time-of-flight cameras, which are connected to the main-controller.

The crawling through bush and briar, the walking over park and pale requires a robust and reliable communication between the host controller, the sensors, and the actuators even in harsh environments. CAN suits these requirements.