ASIMO ASIMO Home
P3-Prototype3
HOME Specifications Technology Movies 3D View
P3
Studies on Leg/Foot Functions of the Robot
Movements of leg/foot joints when walking
We have revealed that the absence of toes has no significant effect on walking. More significantly, support is ensured by the base sections of the toes, i.e., balls of feet, and joint areas. Without the foot joints, one cannot feel contact with the walking surface. Therefore,one is not only vulnerable to back-and-forth instability, but also less stable when crossing diagonally over an inclined surface. It is also impossible to ascend and descend stairs without the knee joints. The absence of coxae makes autonomous walking extremely difficult. A s a result of our examinations, we opted for integrating coxae, knee joints and articulationpedis in our humanoid robot.
Joint alignment
Joint alignment was determined to make it "equivalent" to the human skeletal structure.
Movable ranges of joints
The movable ranges of joints while walking were defined in accordance with walking test measurements on flat surfaces and stairs.
Dimensions, weights and centers of gravity of each leg and foot part
foot part The center of gravity of each part was determined by referring to that of the human body.
Torque application to the joints when walking
Torque acting on the joints was optimized based on the measurements of human joint movement during walking and reaction vectors from the contact surface.
Sensor systems required for walking
Our sense of equilibrium is ensured by three sensing mechanisms. Detection of acceleration is provided by the statoliths. Three semicircular canals detect angular velocity. The bathyesthe sia of muscles and skin is responsible for detecting angles, angular velocity, muscular dynamism, pressures on plantae and sense of contact. Also important is our visual sense, which supports and sometimes compensates for the sense of equilibrium. It also provides information required for normal walking. Thus, our robot system needed to incorporate G-force and six-axial force sensors to detect the conditions of legs/feet while walking, and an inclinometer and joint-angle sensors to detect the overall posture.
Landing impact when walking
A human eases the impact of walking with a combination of structures and functions of movement. The former includes soft skin, ankles and arch-like structures comprised of several bone s at the toe joints. The latter is ensured by the bending motions at joints when the plantae come into contact with the walking surface. Studies of human walking disclosed that the reaction from the surface tends to increase along with an increase in walking speed, even with the above-mentioned shock-easing functions. When walking at a speed of 2-4km/h, the load to theleg/foot is 1.2-1.4 times greater than the body weight. At a speed of 8km/h, the reaction lo ad exceeds 1.8 times the weight. Although the robot had to feature similar shock-absorption mechanisms, the structural measurements were not viable because they might have reduced the robot's stability. Impact absorption was thus ensured through precise control of each component. Based on the results of analysis, we determined the specifications for the robot's leg s and feet.
Development of Two-legged Robot
Based on a prototype design, which was manufactured for t he leg/foot function studies, we established that three functions had to be satisfied: Walking speed corresponding to that of a human (i.e., 3km/h). Provision of structural support for the upper mechanisms, namely arms and hands. Capabilityto go up and down ordinary stairs. The program began with "static walking*l," which was later shifted to "dynamic walking*2," a dominant style of human walking. Naturally, the walking program for the robot was developed according to data on human walking. The active walking program was then integrated into the robot's management system. The continuous study program gradually allowed us to determine specifications. We followed this up by ensuring stable dynamic two-leg/feetwalking operations and, finally, autonomous two-leg/feet walking. (*1) Static walking: The center of gravity is maintained within the supporting leg base area. A smaller footstep and slow speed. (*2) Dynamic walking: The center of gravity is outside the supporting leg base area. A walking maneuver where static balance is intentionally terminated.
For Freer Walking
For the basis of two-leg/foot operation, specifications were determined for straightforward dynamic movement on a flat surface. The next logical step was to conduct a research and development program for freer walking. The robot developed in the following stages had to be capable of walking over undulations and bumps, inclined surfaces and stairs, as well as more stable autonomous walking without the risk of falling. Technical challenges for ensuring robot stability focused on the following three factors:
1
2
3
Technical Clues to Stable Walking
Humans attempt to restore their posture by applying pressure to a part (or parts) of the plantae to avoid falling while walking and at a standstill. If they judge that the pressure application i s not enough, they can change the location of the center of gravity by a structural movement and/or by stepping out. A similar action was required for the robot in order to maintain posture stabilization.
Basic Principles for Posture Control
The robot was basically controlled to follow the joint angle to meet a specific walking pattern. The resultant force from the inertia and gravity determined by this pattern is called "target overall inertia." The point where the overall inertia target becomes zero is defined as "target ZMP." The combined force between the reaction from the floor to the left and right plantae is called "actual overall floor reaction." The point where the actual overall floor reaction is at zero is regarded as a "central point of the actual overall floor reaction." The target ZMP coincides with the central point of the actual overall floor reaction as long as the robot is walking in an ideal manner. The two points, however, tend to differ from each other during actual walking because of the effects of bumps and inclination, even though the upper bodyposture conforms to the target and the joint angle is exactly controlled according to the target. In such conditions, there is a difference in the line of action that results from the floor reaction a nd point of action of the overall target inertia. This leads to coup le generation, which acts on the robot by inclining it. The coupling of the forces is defined as a "moment of falling force," one of the most difficult hurdles to ensuring stable walking control. Hon da's strategy is based on an innovative idea of making the best u se of the falling force moment. Specifically, the difference in the target ZMP and central point of the actual overall floor reaction isdynamically controlled through the use of falling force moment. The body inclination is thus compensated by the falling force moment, which is typically a problematic factor in posture control.
Evolution to Humanoid Robot
Last but not least, Honda engineers had to integrate the leg/foot mechanism. The robot had towalk in a stable and autonomous manner with the upper body to achieve a humanoid design.The functions of Honda's humanoid robot are defined as follows: An operational system that au tonomously performs typical operations under known circumstances. If an extraordinary operation is required under unknown circumstances, the robot will be supported by an operator. The first prototype, P1, had a height of 1,915mm with a weight of 175kg. The P1 was developed to identify the most appropriate way to ensure synchronous arm/leg movements. As such, the prototype does not feature an integrated power source. With the P1, we achieved basic functions,including turning switches on/off, grasping a doorknob and transporting objects held by two hands. Also achieved were the synchronized movements between the arms and legs. The P1 was followed by the humanoid-type autonomous prototype, P2, which adopted wireless techniques. The P2, 1,820mm in height and 210kg in weight, features a computer unit, motor-drive system, battery and wireless apparatus inside the body section. This more sophisticated robot can achieve freer movement, go up and down stairs and push a vehicle. All these functions are trigger ed via a wireless transponder. Naturally, the level of autonomous operation is greater with the P2 than with the Pl. The latest version of the prototype robot is the P3, which was completed in September 1997. Efforts for downsizing resulted in a more compact and lighter design with a height of 1,600mm and a weight of 130kg.
Future Development
In terms of hardware, the program in the future will focus on:
1
2
3
For items 2 and 3, it is extremely important that through the evolution of hardware we achieve physical autonomy by improving dynamic performance and adaptability to wider variations of working conditions. Also important is the pursuit of studies in artificial intelligence systems, which will provide the solution for improved autonomy. If all these are achieved, the robot will not require the support of a human operator for minute correction operations. In terms of software, we should aim at promoting a social infrastructure where humanoid robots will be widely and easily accepted. This is a particularly significant issue when considering the appearance of the humanoid robot. Honda hopes that the time will come when humanoid robots play an important role in serving us and enriching our lives and society.
Page Top

Page Top

Honda Worldwide site

Home | Site Map | Site Index | About this Site

Copyright, Honda Motor Co., Ltd. and its subsidiaries and affiliates. All Rights Reserved.