The Four-Legged Walking Machine AMOS-WD02

The AMOS-WD02 [125] consists of four identical legs. Each leg has two joints (two degrees of freedom (DOF)), which are a minimum requirement to obtain the locomotion of a walking machine and which follow the basic principle of movement of a salamander leg (cf. Sect. 2.2). The upper joint of the legs, called the thoracic joint, can move the leg forward (protraction) and back-ward (retraction), and the lower one, called the basal joint, can move it up (elevation) and down (depression) [12] (Fig. 4.14).

The length of the levers which are attached to the basal joints is propor-tional to the dimension of the machine. They are kept short to avoid greater torque in the actuators [163]. The configuration of the leg, built from a con-struction kit [33], is shown in Fig. 4.13.

58 4 Physical Sensors and Walking Machine Platforms

Basal joint

Lever

Thoracic joint

Fig. 4.13. The leg with two DOF. Left: The 3D model of the leg. Right: The physical leg of the AMOS-WD02

Inspired by vertebrate morphology of the salamander’s trunk and its mo-tion (described in Sect. 2.2), the robot was constructed with a backbone joint which can rotate around a vertical axis. It facilitates a more flexible and faster motion.³The backbone joint is also used to connect the trunk, where two hind legs are attached, with the head, where two forelegs are installed. The trunk and the head are formed with the maximum symmetry to keep the machine balanced for stability while walking. They are also designed to be as narrow as possible to ensure optimal torque from the supporting legs to the center line of the trunk. The construction of the walking machine together with the working space of the legs and the active backbone joint is shown in Fig. 4.14.

The detail of the dimension is presented in Appendix A.

Moreover, a tail with two DOF rotating in the horizontal (y-axis) and vertical (z-axis) axes was implemented on the back of the trunk. In fact, this actively moveable tail, which can be manually controlled, is used only to install a mini wireless camera for monitoring the environment while the machine is walking. However, the tail also gives the walking machine a more animal-like appearance, e.g., in analogy to a scorpion’s tail with its sting [4] (Fig. 4.15).

All leg joints are driven by analog modelcraft servomotors producing a torque between 70 and 90 Ncm. The backbone joint is driven by a digital ser-vomotor with a torque between 200 and 220 Ncm. For the tail joints, micro-analog servomotors with a torque around 20 Ncm were selected. The height of the walking machine is 14 cm without its tail, and the weight of the fully equipped machine (including 11 servomotors, all electronic components, bat-tery packs and a mobile processor) is approximately 3.3 kg. In addition, this machine has two antenna-like sensors and two auditory sensors to perform

3 A walking speed is approximately 12.7 cm/s when the backbone joint is inacti-vated, while it is approximately 16.3 cm/s with the activation of the backbone joint in accordance with the walking pattern. The measurements were done with the walking frequency of the machine at 0.8 Hz.

Working space of the backbone joint Working space of the thoracic joints

Working space of the basal joints

Leg 3 Leg 1

Leg 4 Leg 2

Leg 1

Leg 2

Leg 3, 4 Protraction

Retraction

Retraction Protraction

Elevation

Depression

(a) (b) (c)

Y X

Fig. 4.14. (a) The angle range of the backbone joint (top view). (b) The angle ranges of all thoracic joints on the right side of the walking machine with the left side being symmetric (top view). (c) The angle range of the basal joint of the left foreleg with the remaining legs having the same angle ranges (front view)

X Y

Z Y Y

Z Z

Fig. 4.15. Left: A scorpion’s tail with a sting (modified from S.R. Petersen 2005 [160] with permission). Middle: The tail of the four-legged walking machine AMOS-WD02. Right: The tail of the six-legged walking machine AMOS-WD06. The two DOF tail is constructed in an abstract form of a scorpion’s tail. It is mainly used to install the camera

different reactive behaviors; e.g., an obstacle avoidance and a sound tropism, respectively. The 3D model of the walking machine and the real walking ma-chine are shown in Fig. 4.16.

60 4 Physical Sensors and Walking Machine Platforms

Wireless camera

Active tail

Auditory sensors

Backbone joint

Antenna-like sensors

Fig. 4.16.The four-legged walking machine AMOS-WD02.Left: The 3D model of the walking machine.Right: The real walking machine

All in all, the AMOS-WD02 has 11 active DOF, 4 sensors and 1 wireless camera (for more details of the AMOS-WD02, see Appendix A). Therefore, it can serve as a reasonably complex platform for experiments concerning the function of the neural perception–action systems.

However, to test the neural controller and to observe the resulting behav-ior of the walking machine (e.g., obstacle avoidance), they were first simulated in the physical simulation environment YARS (cf. Sect. 3.3). The simulator, developed at the Fraunhofer Institute in Sankt Augustin, is based on Open Dynamics Engine (ODE) [189]. It provides a defined set of geometries, joints, motors and sensors which is adequate to create the four-legged walking ma-chine AMOS-WD02 with IR sensors in a virtual environment with obstacles (Fig. 4.17).

The YARS enables first implementation which is precise enough to re-produce the behavior of the physical walking machine with sufficient quality.

This simulation environment is also connected to the ISEE, which is a software platform for developing neural controllers (described in Chap. 3).

In the final stage, a neural controller which is developed after the test on the simulator is then applied to the physical walking machine to demonstrate the behavior in the real environment. The neural controller is programmed into a mobile processor (a PDA). The PDA is interfaced with the MBoard, which digitizes sensory signals and generates a pulse width modulation (PWM) sig-nal at a period of 20 ms, to command the servomotors. The communication between the PDA and the MBoard is accomplished via an RS232 interface at 57.6 kbits/s.

Fig. 4.17. Different views of the simulated walking machine in its environment.

The properties of all simulated components are defined with respect to the physical properties of the real walking machine, e.g., weight, dimension, motor torque and so on. The simulated walking machine consists of body parts (head, backbone joint, trunk and limbs), servomotors and IR sensors, while the auditory sensor was not available in the simulation