Gill A. Pratt

Virtual Model Control: An Intuitive Approach for Bipedal Locomotion

Virtual model control is a motion control framework that uses virtual components to create virtual forces generated when the virtual components interact with a robot system. An algorithm derived based on the virtual model control framework is applied to a physical planar bipedal robot. It uses a simple set of virtual components that allows the robot to walk successfully over level terrain. This paper also describes how the algorithm can be augmented for rough terrain walking based on geometric consideration. The resulting algorithm is very simple and does not require the biped to have an extensive sensory system. The robot does not know the slope gradients and transition locations in advance. The ground is detected using foot contact switches. Using the algorithm, we have successfully compelled a simulated seven-link planar biped to walk blindly up and down slopes and over rolling terrain.

Series elastic actuator development for a biomimetic walking robot

Series elastic actuators have linear springs intentionally placed in series between the motor and actuator output. The spring strain is measured to get an accurate estimate of force. A second order linear actuator model is broken into two fundamental cases: fixed load-high force (forward transfer function), and free load-zero force (impedance). This model is presented with dimensional analysis and extends previous linear models to include friction. Using the model and dimensionless groups, we examine nonlinear effects of motor saturation as it relates to large force bandwidth and nonlinear friction effects such as stiction. The model also helps to clarify how the springs help and hinder the operation of the actuator. The information gained from the model helps to create a design procedure for series elastic actuators. Particular emphasis is placed on choosing the spring constant for the elastic element.

Virtual model control of a bipedal walking robot

The transformation from high level task specification to low level motion control is a fundamental issue in sensorimotor control in animals and robots. This paper describes a control scheme called virtual model control that addresses this issue. Virtual model control is a motion control language that uses simulations of imagined mechanical components to create forces, which are applied through real joint torques, thereby creating the illusion that the virtual components are connected to the robot. Due to the intuitive nature of this technique, designing a virtual model controller requires the same skills as designing the mechanism itself. A high level control system can be cascaded with the low level virtual model controller to modulate the parameters of the virtual mechanisms. Discrete commands from the high level controller would then result in fluid motion. Virtual model control has been applied to a physical bipedal walking robot. A simple algorithm utilizing a simple set of virtual components has successfully compelled the robot to walk continuously over level terrain.

Intuitive control of a planar bipedal walking robot

Bipedal robots are difficult to analyze mathematically. However, successful control strategies can be discovered using simple physical intuition and can be described in simple terms. Five things have to happen for a planar bipedal robot to walk. Height has to be stabilized. Pitch has to be stabilized. Speed has to be stabilized. The swing leg has to move so that the feet are in locations which allow for the stability of height, pitch, and speed. Finally, transitions from support leg to support leg must occur at appropriate times. If these five objectives are achieved, the robot will walk. A number of different intuitive control strategies can be used to achieve each of these five objectives. Further, each strategy can be implemented in a variety of ways. We present several strategies for each objective which we have implemented on a bipedal walking robot. Using these simple intuitive strategies, we have compelled a seven link planar bipedal robot, called Spring Flamingo, to walk. The robot walks both slowly and quickly, walks over moderate obstacles, starts, and stops.

Stiffness Isn't Everything

Most robot designers make the mechanical interface between an actuator and its load as stiff as possible[9][10]. This makes sense in traditional position-controlled systems, because high interface stiffness maximizes bandwidth and, for non-collocated control, reduces instability. However, lower interface stiffness has advantages as well, including greater shock tolerance, lower reflected inertia, more accurate and stable force control, less damage during inadvertent contact, and the potential for energy storage. The ability of series elasticity (usually in the form of a compliant coating on an end-effector) to stabilize force control during intermittent contact with hard surfaces is well known. This paper proposes that for natural tasks where small-motion bandwidth is not of paramount concern, actuator to load interfaces should be significantly less stiff than in most present designs. Furthermore, by purposefully placing the majority of interface elasticity inside of an actuator package, a new type of actuator is created with performance characteristics more suited to the natural world. Despite common intuition, such a series-elastic actuator is not difficult to control.

Force controllable hydro-elastic actuator

We present a hydro-elastic actuator that has a linear spring intentionally placed in series between the hydraulic piston and actuator output. The spring strain is measured for an accurate estimate of force. This measurement alone is used in PI feedback to control the force in the actuator. The spring allows for high force fidelity, good force control, minimum impedance, and large dynamic range. A third order linear actuator model is divided into two fundamental cases: fixed load-high force (forward transfer function), and free load-zero force (impedance). These two equations completely describe the linear characteristics of the actuator. This model is presented with dimensional analysis to allow for generalization. A prototype actuator that demonstrates force control and low impedance is also presented. Dynamic analysis of the prototype actuator correlates well with the linear mathematical model.

Legged robots at MIT: what's new since Raibert?

The MIT Leg Lab is best known for the seminal work of Marc Raibert, who showed in the 1980s that robotic running could be accomplished using a few simple, decoupled control laws. Raibert had shown how to calculate the control effort required for the next step of a run based on the results of the previous step, using simple physics-based rules. But smoothly controlling joints in a continuous-time fashion to effect walking seemed more difficult, particularly for non-stead state conditions. It was unclear how to combine passive dynamics with active control. Other researchers were beginning to address robot walking, and the rich diversity of their approaches was very exciting. It thus seemed that walking was the next important area. We embarked on a two-pronged approach: one focusing on making improvements to electromechanical actuators and robot design, the other on walking algorithms and control.

Virtual actuator control

Robots typically have an individual actuator at each joint which can result in a nonintuitive and difficult control problem. In this paper we present a control method in which the real joint actuators are used to mimic virtual actuators which can be more intuitive and hence make the control problem more straightforward. Our virtual actuator control method requires a solution to the force distribution problem when applied to parallel mechanisms. An extension of Gardners partitioned actuator set control method (1991) is presented. This extended method allows for dealing with constrained degrees of freedom in which the torque cannot be specified but can be measured. A simulated hexapod robot was developed to test the proposed control method. The virtual actuators allowed textbook control solutions to be used in controlling this highly nonlinear, parallel mechanism. Using a simple linear control law, the robot walked while simultaneously balancing a pendulum and tracking an object.

Dynamic bipedal walking assisted by learning

This paper presents a general control architecture for bipedal walking which is based on a divide-and-conquer approach. Based on the architecture, the sagittal-plane motion-control algorithm is formulated using a control approach known as Virtual Model Control. A reinforcment learning algorithm is used to learn the key parameter of the swing leg control task so that speed control can be achieved. The control algorithm is applied to two simulated bipedal robots. The simulation analyses demonstrate that the local speed control mechanism based on the stance ankle is effective in reducing the learning time. The algorithm is also demonstrated to be general in that it is applicable across bipedal robots that have different length and mass parameters.

The DARPA Robotics Challenge [Competitions]

ny news broadcast reveals that Mother Nature and human nature are not always on our side. Shifting demographics make society increasingly susceptible to both natural and man-made disasters. Densely populated coastal cities prone to flooding and bad weather invite heavy tolls from natural disasters. Manmade disasters come in many forms, such as chemical spills, weapons of mass destruction, release of radiation, and so on. Humanity must improve its defenses against such threats, i.e., we must prepare for the unexpected. During the Fukushima disaster, many Japanese citizens, aware of the development of humanoid robotic platforms, asked, “Can’t robots help us?” It turns out that they could not, at least not then. The Defense Advanced Research Projects Agency (DARPA) Robotics Challenge (DRC) aims to work up to robots that can help in the near future. The DRC is a program in human-scaled robotics for disaster response. DARPA’s goal is to create robots that could interface with human environments, use human tools, and be commanded by humans without specialized training. We want robots that can adapt to assist in any disaster.