For the fast movement in mammalian, quadruped animals such as rat, horse or dog are able to make certain velocities in uneven terrains. These advantages have attracted much attention in engineering area in developing the legged robot systems. From the beginning in 1960s, the structure of the robot system con-sists of a single rigid body for the trunk and the four legs. With these components, many researches have been shown from the gait pattern control to stabilization of its posture.
However, these mentioned robotic systems have had a little attention to the movement of the body trunk. In nature, feline mammalian such as cheetah makes nearly 120 [km/h]. However, few legged robots have reached the maximum speed of the living feline animals. Although it is possible to find the reason in many aspects in mechanical point of views such as its balance, speed, friction and the step cycle by leg combination, the relation between the spinal and pelvic motion during the running behavior has not extensively researched in this area. As a onsequence, the aim of this paper is to develop and analyze the spinal motion mechanism for the bio-inspired quadruped robotic system.
In this study, the relation of the spinal and pelvic motion during the galloping gait is firstly examined by using a real feline animal (Felis Catus Domestica). In the experimental design, we set 10 fluorescent mark-ers according to the leg joints and configurations of spinal column. Then four galloping sections were ob-served with six phases from low and high speed galloping. Based on thorough analysis, the spinal motion coincided with pelvic motion with delayed angular phase at the hindlimb stance phase during low speed galloping. Due to this delay, the speed of the body stayed at the 3.0m/s region. Compared to low speed galloping, the spinal motion was matched with the pelvic motion so that the speed was increased from 3.20m/s to 4.33m/s in high-speed galloping. Based on these observations, the association between spinal and pelvic motion match at the hindlimb stance may increase the ground reaction force so that the ballistic flight of the body is generated. Therefore, the relative spinal motion synchronization is a key fac-tor for the speed increase during the galloping gait.
With the result from the observation of the spinal motion, this study proposes the planar robotic platform called ‘Sprinter’, containing the spinal mechanism on the body of the platform. In terms of the mechanism of the quadruped gait, a system is built on the 4 under-actuated legs and the body trunk having an active spine joint. For the consideration of the planar movement with tethered boom, this platform moves both the flat treadmill and track environments.
Based on the bio-inspired platform, the role of the spinal motion is verified with the optimal spinal mo-tion profile generation. In order to do this, the Q-learning algorithm was applied to the kinematic model of the forelimb and hindlimb stance phase. After the spinal motion profile re-design, this profile is testified by the dynamic simulator and the 2D-planar experimental set. Based on the experiments, the re-designed spinal profile encourages the bigger ground reaction force and pivot effect of the center of mass with higher speed increase in the unit bounding gait.