Nicolas Lauzier

A flexible robot skin for safe physical human robot interaction

Providing contact sensing on the whole body of a robot is a key feature to increase the safety level of physical human-robot interaction. In this paper, a new robot skin capable of sensing multiple contact locations is presented. The motivation behind the proposed design is to produce a relatively inexpensive skin having the capability to provide the spatial location of collisions and also to add compliance to the robots external cover. The resulting device is a thin flexible sensor sheet made of polyimide films with electrically conductive ink and a pressure sensitive conductive rubber sheet. The problem of internal wire routing is circumvented by the use of conductive ink and a circuit is proposed to minimize the number of output wires. To provide collision absorption and mechanical robustness, the sensor is embedded in different layers of polyurethane using shape deposition manufacturing (SDM). The paper presents the design and the fabrication process of the skin but also some experimental results on the determination of the mechanical properties of the resulting sensor as well as its potential for increasing human safety during human robot interaction.

Development of a Compact Magnetorheological Fluid Clutch for Human-Friendly Actuator

There is a strong demand for a human–machine-coexistent machine, e.g., a power-assist system or a computer-aided rehabilitation system, in the context of the super-aging society. In such a human–machine-coexistent system, it is important to guarantee hardware-level safety by utilizing human-friendly actuators. In this study, we have developed a compact magnetorheological fluid clutch (CMRFC) for human-friendly actuators. In this paper, at first, the basic design of the CMRFC and developments of 5- and 40-Nm class devices are explained. Then we discuss experimental results on the basic characteristics of these devices. In the final part, a new magnetorheological fluid that includes nano-sized particles is discussed.

Series Clutch Actuators for safe physical human-robot interaction

This paper presents the design, implementation and control of a device intented to mechanically improve the safety of serial robots interacting with humans. The device consists of an electronically adjustable torque limiter placed in series with each actuator, referred to as a Series Clutch Actuator (SCA). By appropriately adjusting the limit torques according to the robots configuration, the maximum static force that the robot can apply to its environment at the Tool Centre Point (TCP) can be limited to a prescribed safe level. If a limit torque is exceeded, the SCA slips and an emergency stop is triggered while the inertia located upstream from the SCA in the kinematic chain is mechanically disconnected. A method is presented to determine the optimal limit torques that maximize the isotropically achievable force (which can be applied in all directions without triggering any SCA) while satisfying the safe force limit. An approach to optimize the pose of a redundant robot in order to maximize the isotropically achievable force while preserving a safe maximum force threshold is also proposed. The design and fabrication of a torque limiter using a large number of friction discs is presented. Finally, the mechanisms are implemented into a 4-DOF redundant serial arm and preliminary experimental results are presented.

3-DOF Cartesian Force Limiting Device Based on the Delta architecture for safe physical human-robot interaction

This paper presents a device that significantly improves the safety of ceiling-mounted robots whose end effector orientation remains constant with respect to the vertical direction (e.g. Scara-type robots). The device consists of a three-degree-of-freedom (DOF) parallel mechanism with the Delta architecture on which the revolute actuators have been replaced with torque limiters. The resulting Cartesian force limiting device (CFLD) is implemented as a mechanical connection between the robot and the effector. It is rigid unless excessive forces are applied on the end effector, for example during a collision. The magnitude of force that activates the mechanism is set by properly adjusting the threshold of the torque limiters. Furthermore, a collision can be rapidly detected with a limit switch placed on one of the links of the mechanism and a signal can be sent directly to brakes that will stop the robot, without passing through a controller and thus improving the reliability and reaction-time of the safety system. By mechanically disconnecting the robot from its end effector, the device ensures that the person involved in the collision is only subjected to the inertia of the end effector and thus potential injuries are greatly reduced. This work is the extension of a previous 2-DOF CFLD that was sensitive only to horizontal forces. The new architecture reacts to collisions occuring in any direction and is geometrically optimized for the proposed application.

2 DOF cartesian force limiting device for safe physical human-robot interaction

This paper presents a device that significantly increases the safety level of suspended robots whose end-effector orientation remains constant with respect to the vertical direction (e.g. Scara-type suspended robots). The device is a two-degree-of-freedom (DOF) parallel mechanism with a parallepipedic architecture on which two revolute joints have been replaced with commercially available torque limiters. The device is implemented as a mechanical connection between the robot and the effector. It is rigid unless excessive horizontal forces are applied on the end-effector, for example during a collision. The level of force that activates the mechanism is set by properly adjusting the threshold of the torque limiters. Furthermore, a collision can be rapidly detected with a limit switch placed on one of the links of the mechanism and a signal can be sent directly to brakes that will stop the robot, without passing through a controller and thus improving the reliability and reaction-time of the safety system. By mechanically disconnecting the robot from its end-effector, the device ensures that the person involved in the collision is only subjected to the inertia of the end-effector and thus potential injuries are greatly reduced. A prototype of the proposed device has been built to validate the concept and to study its behaviour for collisions with different velocities and orientations.

Performance Indices for Collaborative Serial Robots With Optimally Adjusted Series Clutch Actuators

Safety is the first priority when designing robots that are intended to physically interact with humans. New robotics standards state as a condition for collaboration that the robot should be designed so that it cannot exert forces larger than 150[N] at its tool centre point. An effective and reliable way of guaranteeing that this force cannot be exceeded is to place a torque limiter in series with each actuator, thus forming Series Clutch Actuators (SCAs). Since the relationship between the joint limit torques and the achievable end-effector forces is configuration dependent, it is preferable to use adjustable torque limiters. This paper presents a method to optimally control the limit torques of a serial manipulator equipped with adjustable series clutch actuators. It also introduces two performance indices to evaluate the quality of the relationship between the joint limit torques and the achievable end-effector forces. The first one is the ratio of the minimum and maximum force thresholds. Even if it has a strong physical meaning, it is not differentiable everywhere in the workspace and is thus difficult to use in an optimization process based on its gradient. A second index, smooth and expressed in a closed-form, is therefore introduced which is the determinant of the normalized Jacobian matrix post-multiplied by its transposed. Examples of redundant manipulator motion optimization and of collaborative robot architecture optimization using the second index are shown. Finally, the main limitations of the proposed methods and indices are discussed.Copyright

Adaptive gravity compensation of decoupled parallel and serial manipulators using a passive hydraulic transmission

Abstract This article addresses the gravity compensation of decoupled or partially decoupled parallel and serial manipulators. The proposed technique applies to manipulators in which the vertical motion is decoupled from other Cartesian degrees of freedom (DOFs). In the latter situation, only one DOF needs to be gravity compensated in order to eliminate actuator torques due to the weight of the moving parts. First, a passive hydraulic transmission consisting of two cylinders connected by hoses is introduced in order to perform gravity compensation using a remote counterweight (CW) affecting only the inertia of the equilibrated DOF. A complete system is then shown using a set of CWs, cylinders, and valves allowing to adapt rapidly and effectively to a variety of payloads. Finally, two experimental implementations using, respectively, hydraulic and pneumatic cylinders filled with water are presented and the practical limitations of the approach are discussed.

On the Design of Human-Safe Robot Manipulators

Bringing robot manipulators in the same environment as humans seems a natural evolution in the path towards more advanced robotics. This upcoming co-existence will offer a tremendous potential to improve many industrial applications such as manufacturing and assembly. In this paradigm, an efficient synergy between human and robot can be obtained by combining the human’s reasoning ability and adaptability in unstructured environments with the inexhaustible strength of robots. The current generation of commercially available robot manipulators is not designed to fit the specific needs required by this novel collaboration. Indeed, control algorithms that enable an intuitive and efficient interaction between humans and robots are still missing to industrial robots. At an even more fundamental level, the way they are currently designed presents significant risks in the proximity of humans. Many studies have investigated this last aspect to demonstrate the potential danger of a robot Zinn, Khatib, Roth & Salisbury (2004a) and to understand and provide metrics to characterize to the level of this threat Haddadin, AlbuSchaffer, Frommberger & Hirzinger (2008); Haddadin, Albu-Schaffer & Hirzinger (2008); Yamada et al. (1997). The next step for robot designers should focus on increasing human safety to an acceptable level according to the conclusion of these studies. The aim of this chapter is to present how, at the conceptual level, robot manipulators should be mechanically designed to be harmless for humans. Both established and novel concepts will be reviewed to provide actual guidelines to the robot designer. Serial elastic actuators (SAE), distributed macro-mini (DM2) and variable stiffness joints will be reviewed whereas more emphasis will be placed on force limiting devices (FLD), robot soft covering and a method for efficiently coupling robot joint actuators for reducing their potential of transferring energy to the surrounding environment.

Towards Whole-Arm Statics

In this paper, a novel approach is proposed to compute the maximum contact force that can be applied by a manipulator at any point on its external surface. The approach consists in the approximation of the robot’s external surface by simple primitives for which the position and normal direction can easily be expressed analytically for any point on the surface. This approach is also used to compute analytically the boundaries of the zones on the external surface where the achievable force exceeds a predetermined threshold. A method is developed for determining the limit joint torques which ensure that the achievable force cannot exceed a predetermined magnitude for any point on the primitives. Specific equations are developed for two types of primitives (parallelogram and cylinder) along with generic procedures applicable to any shape. An example is presented to illustrate how the proposed procedures can be applied for the solution of a typical problem. Finally, the applicability and limitations of this approach are discussed. A potential application for this work is to provide information on a robot’s dangerousness to its controller in order to improve safety.Copyright