Mehdi Benallegue

The Yoyo-Man

The Yoyo-Man project is a research action tending to explore the synergies of anthropomorphic locomotion. The seminal hypothesis is to consider the wheel as a plausible model of bipedal walking. In this paper we report on preliminary results developed along three perspectives combining biomechanics, neurophysiology and robotics. From a motion capture data basis of human walkers we first identify the center of mass (CoM) as a geometric center from which the motions of the feet are organized. Then we show how rimless wheels that model most passive walkers are better controlled when equipped with a stabilized mass on top of them. CoM and head play complementary roles that define what we call the Yoyo-Man.

Metastability for High-Dimensional Walking Systems on Stochastically Rough Terrain

—Biped walking is a complex task, but usually with a natural limit-cycle behavior when walking on an even ground. However, perturbations during walking can make the robot fall. Several works addressed the issue of measuring the robustness to disturbances, and most methods study the effect of a single perturbation. But when walking, the disturbances can be multiple, such as walking on rough terrains. The metastability is a concept that helps studying the case of multiple disturbances. The performance measure is the expectation of the time during which the robot can keep balance. However, until today, only two methods permit to measure this metrics: the discretization of all the state-space and the Monte-Carlo sampling. The former one cannot address high dimensional state-space and the latter is too much time-consuming when the falls are rare. We propose here a solution for walkers with high-dimensional states, even when falls are very rares. The novelty of this method is to rely on the property that limit-cycle walkers may return to the limit-cycle several times before to fall. This method is even extended to the cases of bifurcations or chaos. We illustrate the performance of the approach with a simulated high-dimensional actuated walking system.

Impact acceleration of falling humanoid robot with an airbag

Maximum absolute acceleration at impact is widely used for evaluating damage at robot fall. In this paper, we measured the impact acceleration of a life-size humanoid robot when it fell down. The experiment showed that a robot suffered maximum acceleration of over 100G at the impact. To mitigate this we tested an airbag system. Our experiment showed the airbag could reduce the peak acceleration to between 20G and 30G, which is an acceptable level for the most of robot hardware.

Bipedal Locomotion: A Continuous Tradeoff Between Robustness and Energy-Efficiency

Walking is a mechanical process involving all the limbs the body and subject to balance constraints. The structure of the body can allow the emergence of a stable walking using few to no energy, but this motion is sensitive to external perturbations and prone to fall easily. On the other hand, with strong actuation and anticipation, much more disturbances can be overcome, but at the cost of high energy consumption. The choice between these two modes of locomotion is more than a context-dependent binary selection. It is a continuous tradeoff, not only providing a span of different walking control policies but building the framework of a precise classification of these controllers. In this chapter, we discuss, for the humans and the robots, the two extreme walking paradigms and how to sort the solutions ranging between them. We analyze also the incentives behind the decision to change to a more robust or more efficient control policy. Finally, we show that the versatility of a walker depends deeply on the relationships lying between its controller dynamics and its mechanical design, both being ideally built together in a process of codesign.

Robot Motion Planning and Control: Is It More than a Technological Problem?

The generation of motion for robots obeys classically to a two-step paradigm. The first step is the planning, where the typical problem is to find a geometric path that allows the robot to reach the desired configuration starting from the current position while ensuring obstacle avoidance and enforcing the satisfaction of kinematic constraints. Motion planning lays its grounding on the decidability properties of this classic geometrical problem. Moreover, the traditional approaches that are used to find solutions rely on the global probabilistic certainty of the convergence of path construction stochastically sampled in the configuration-space. The second step of motion generation is the control, where the robot has to perform the planned motion while ensuring the respect of dynamical constraints. Motion control seeks primarily for local controllability or at least the stability of the motion. The basic instances of this problems have long been tackled using local state-space control. However, the typical nonlinearity of the dynamics, together with the non controllability of its linearization, lead more and more solutions to resort to model predictive control. These methods make it possible to predict the outcome of a control strategy in a future horizon and to improve it accordingly, commonly by using numerical optimizations which take into account the safety constraints and efficiency intents. However, since few years, the improvement of computational capabilities and numerical algorithms allows more and more to deal with complex dynamical systems and for longer horizons. This allows then these approaches to untighten the local nature of their applications and progressively start wider explorations of their reachable space. This evolution brings us to the question of the rising overlap between planning and control. Today, most planning problems would take too much time to be solved online with numerical approaches. Does that imply that the generation of motion will theoretically never be free of the necessity of a prior planning? Or on the contrary, is planning only a numerical issue?