In this part of the site we are going to see what are the interests of walking for a robot, the constraints and the solutions provided. Concretely do we manage to make robots work today?
Many laboratories are working on the subject and in particular the “Leg Laboratory” of MIT. However, today walking is far from being mastered. We manage to make people dance, descend or climb stairs but an obvious lack of ease still exists in robots, whether they are bipedal or quadruped. First question: Why walking?
The advantage of making a walking robot allows it to move over a wider variety of terrain than a robot roller for example allows. The ground can thus be irregular or even strewn with small obstacles. So the final goal would be to be able to use the robot in any place initially accessible by humans. Certain problems nevertheless arise for the implementation of walking. In particular the obvious problem of balance. Managing the robot’s center of gravity then becomes essential. Two solutions can be proposed to solve this problem: Have a low center of gravity A wide body that makes it easier to maintain stability Here is an example of bipedal robots:
Why do you need a low center of gravity?
We can have a greater inclination (displacement of the center of gravity) of the body when moving the robot since the center of gravity will remain inside the surface delimited by the points of support on the ground (surface delimited by the surface of the feet). So the robot is more likely to stay upright.
There are usually two types of steps:
- Dynamic walking.
- Quasi-static walk.
Dynamic walking corresponds to walking the most similar to that of Man. The problem in this type of walk is essentially the loss of balance between each step. In addition, being able to stop while walking also creates difficulties.
Quasi-static walking is easier to implement because the robot will only very rarely be in a situation of imbalance.
Here is an explanatory diagram of dynamic walking:
The diagram above shows 4 steps resulting from the decomposition of 2 robot steps. The red dot represents the robot’s center of gravity.In the first step (step 2), the robot’s body moving forward, the center of gravity is deviated from its equilibrium point. The phenomenon is also repeated during the second step (step 4). It is therefore because of the displacement of the robot’s center of gravity that it loses its balance and risks falling. This problem is increased by the fact that the 2 steps are carried out successively without an intermediate stop.
Let us now study the quasi-static walk. A first solution consists in using large feet surrounding the central zone of the robot. To do this one can use for example “U” shaped feet. However, a problem arises when changing direction.
On the diagram above we can see the decomposition of 2 successive steps of the robot. We can see here that the center of gravity remains in the equilibrium zone also called a support polygon (in dotted lines on the diagram).
A second quasi-static walking solution allows the use of more human-shaped feet, thus leaving changes of direction easier to carry out. This solution consists in fact in bringing the center of gravity above the support polygon.
To do this, before each step, part of the mass of the robot must be moved above the foot remaining in support.
On the upper part of this diagram, it is about an example of dynamic walk which we spoke previously and which poses problems of imbalance.
On the other hand, on the lower part of the diagram we can see that before taking his step the robot moves the upper part of his body above the balance zone remaining during the step.
To conclude, we can say that static walking can be implemented relatively easily. Walking on the other hand, during
changes of direction induce a large number of motors and the complexity of the assembly increases. However, walking is essential for interaction with the environment (avoiding an obstacle, …).
