|
|
||
|
|||
Humanoid Robot Soccer Tournament |
||||
|
||||
| Legged locomotion is a particularly intriguing challenge in robotics. Legged robots have distinct advantages over their wheeled counterparts when working in arbitrarily complex terrains: they can generally cross obstacles more easily and exhibit greater adaptability. These features are important in several applications such as exploration, maintenance, intervention and service, all of which are beyond the realm of traditional manufacturing robots. Bipeds are especially interesting because humans are two-legged and places meant to be occupied by humans (such as homes and offices) are designed for two legged locomotion. If machines are to become more commonplace phenomena in our lives (which clearly is the trend), the technology of two-legged robotic locomotion in arbitrary environments must be mastered thoroughly. | ||||
|
||||
With a height of 46 cm and a weight of only 2 kg, the RoboSapien is small in size but it has a sophisticated overall mechanical structure. RoboSapien has 17 degrees of freedom (DOF); there are 17 joints powered by motors which must be controlled simultaneously for the robot to even stand up (if un-powered, the structure simply crumples into a heap just like how a bicycle chain would). In order to realize a natural gait, the joint design of the robot is modeled on the structure of the human body, with each 6-DOF leg enabling the robot to imitate most of the human walking motions. The robot senses its environment through a combination of eight force sensors, one tilt sensor, an IR sensor, a digital compass and a video camera. To coordinate its sensory information and walking, the robot uses a two-layered control system. The high-level control layer gathers and processes the information from the sensors, decides on the necessary walking motions and directions, and sends commands to the low-level controller. The low-level control then processes inputs from the tilt and force sensors, and realizes the walking motions and any special motions such as kicking a ball. The competition required each robot to perform 4 different tasks. They are Robot Dash and Backward Walk, Penalty Kick and Goal Keep, Obstacle Run, and Lift and Carry. In the Robot Dash, the robots had to complete a distance of 1.2m using forward bipedal locomotion and then walk backwards in the shortest time possible. Fast walking is very difficult to realize in biped robots, and RoboSapien used a dynamic balance algorithm to improve the walking speed while still maintaining a steady gait. A separate algorithm was also developed to implement RoboSapien's fast back-walking gaits. In this task, our robot managed to set a new world record of 16 seconds, shaving off 10 seconds from the previous record in the forward dash and completed the entire race in 52 seconds. The second task is Penalty Kick and Goal Keep. This task requires the robot to use artificial intelligence techniques so as to locate a randomly-placed ball and the goalpost, generate a trajectory for the ball, and kick the ball according to the trajectory generated. RoboSapien managed to score 1 goal in 5 tries with 1 shot saved by the goalkeeper and another narrowly missing the goal. In the third task, the Obstacle Run, the robots have to overcome two challenges. The first challenge is to walk along a path whilst avoiding obstacles randomly placed in front of the robot. The second challenge is that the robots are also required to remember the paths they have taken. RoboSapien used IR proximity sensors to detect the obstacles, with a servomotor driving the IR sensor to scan laterally, and a digital compass to sense the direction. Lastly, in the Lift and Carry event, the robots were required to carry extra batteries as additional weight while continuing to walk. New walking gaits for the robot were developed for this event and RoboSapien managed to lift a total of 11 AA-size batteries. Based on its performance in all 4 events, RoboSapien brought home the overall 2nd position in the FIRA HUROSOT category.
|
||||
|
 |
