Compliance: An Approach to Physical Interaction in a Kinetic Architecture
The issue of human safety in interactions with robots was considered early in science fiction. American author Isaac Asimov (1950) introduced “The Three Laws of Robotics” in his short story “Runaround” to govern the behaviour of robots:
“ 1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2. A robot must obey the orders given it by human beings, except where such orders would conflict with the First Law.
3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws. ” (Asimov, 1950)
During interactions between robots and humans, the closer the robot and the human are, the more intricate the relationship between robot and human becomes and consequently the more risk of a human being injured by the robot.
To ensure a human’s safety when they have physical contact with a robot, the robot must be equipped with a complete system, that incorporates “(i) safe mechanical design, (ii) human friendly interfaces such as natural language interaction and (iii) safe planning and control strategies” (Kulić and Croft 2005, p383). This blog article focuses on (i) by dealing with the compliance of joints.
The term “compliance” refers to the flexibility of a robot manipulator in industrial robotics. It equals to the value of displacement divide force (Makino, 2014). Based upon this equation of compliance, an apparatus with a high compliance moves considerably for a given force. In other words, the property of compliance allows a robot to adapt to unpredictable contact forces by conducting a modified trajectory of movement. This property leads the compliant robot manipulator to be physically moved by a human, that consequently allows the robot to have a safe interaction with humans when it is in physical contact with a human.
One of the most representative applications of compliant joints is Selective Compliance Articulated Robot Arm, which is often known by its acronym SCARA. The structure of industrial SCARA robots was developed by taking inspiration from the Japanese folding screen, Byobu, as shown in Figure 1. This structure provides SCARA robots a high rotational compliance in the horizontal plane and a high stiffness along vertical axes, leading the SCARA robot to have pliable rotation in the horizontal planes without any vertical motion.
Every arm of a SCARA robot is a non-deformable object that is connected with adjoining arms by a joint system; so to speak, the compliant motion of each arm is achieved by the compliance of joint mechanisms. Therefore, the joint study of examples of existing SCARA robots is essential. The overview of the existing robotic constructions shows that every joint actuation can be implemented in two different ways; namely by active and passive compliance.
“Active compliance systems are computer controlled systems where compliance characteristics can be implemented through software control” (Ahmed, 2011, p16). Generally, compliant motion of joints in active compliance system is achieved by the combination of a speed control system and a sensor based control system. Common speed control systems in existing robotic construction are composed of motors or motors with geared transmission systems. The rotation speed and the position of joints are directly controlled by the speed control systems. The sensor based control systems in active compliance systems involves sensors which can detect the torque or the speed of a joint, or the force applied on a joint. The control system will then process these sensory data received by the sensors and results in a modified movement. A conventional SCARA robot utilises an active compliance system. Each arm of a conventional SCARA robot is individually controlled by a servo motor at each joint. Optical encoders are used to detect the rotational speed and the angular position of arms, and then send feedback signals to servo motor. As a consequence, the servo motor directly controls the rotational speed and the position of arms based on these sensory data received from these optical encoders.
Passive compliance systems are not directly connected with motors. Instead, they are based on a passive mechanism, such as springs. In physical human robot interaction, joints with a passive compliance system can passively adapt to contact forces which are generated by humans. Contrary to active compliance joints, passive compliance joints remain compliant even in the cases of electrical failure (Ahmed, 2011, p21).
Both active compliance joint and passive compliance have pros and cons. A robot manipulator with active compliance joints moves at a given, controlled speed because of the application of servo-control systems at each joint. That means this robot manipulator cannot be physically moved in a physical interaction with humans. Further more, the active compliance system also provides a delayed contact response due to the time needed to process appropriate sensory data (Ahmed 2011). This delayed response would lead to a slow response to the servo motor, and therefore results in an unreliable safety in its physical interaction with a human.
A robot manipulator with passive compliance joints, compared with a robot manipulator with active joints, has more flexible movement. It is able to move freely and adapt to applied forces passively in its physical contact with a human due to the use of non-motorised joints with passive mechanism. However, an uncontrolled inertial motion occurred in a passive compliance system would lead to a potential injury to a human.
In order to achieve a safe physical human robot interaction, a prototype employing both passive joints and active joints, with a clutch and brake system was constructed and tested with a dancer in a performative choreography, see Figure 2.
As illustrated in Figure 3, this prototype of SCARA robot has three limbs and is driven by one central stepper motor. The bottom limb is connected to the stepper motor by a pulley-and-belt transmission, with a clutch and brake system. The clutch is able to disengage the bottom limb from the stepper motor, allowing the bottom limb to converting from an active compliance system to a passive compliance system. The brake can stop the bottom limb when the clutch is disengaged. These three limbs are connected with passive compliant joints, which are consisted of only bearings with brakes. This type of joint mechanism allows the middle and the top limbs to be able to physically moved by a human and generate natural movements due to the kinetic energy transfer between each limb. The brake systems allow these two limbs to stop when they are required, and therefore lead to a safe interaction with huamns.
Passive compliance joints used in the proposed prototype allow limbs to rotate freely and generate natural trajectories. The application of clutch and brakes adds another layer of control to passive compliance joints and consequently ensures humans’ safety in their physical interactions with robots. For more details about mechanisms and structures of this prototype of SCARA robot, please see the Project Trespass.
- Asimov, I. (1950). Runaround. I, Robot. United States: Gnome Press.
- Ahmed, R.M. (2011). Compliance Control of Robot Manipulator for Safe Physical Human Robot Interaction. Örebro Studies in Technology 45. p1-105.
- Kulić, D., Croft, E.A.. (2005). Safe Planning for Human-Robot Interaction. Journal of Robotic Systems. 22 (7), p383-396.
- Makino, H. (1980) Selective Compliance Assembly Robot Arm. 1st International Conference on Assembly Automation (ICAA), p77-86.
- Makino, H. (2007). Review: Development of the SCARA. Journal of Robotics and Mechatronics. Vol.33, p5-8.
- Cycloid-E. (2009). Cod.Act – Cycloid-E. [ONLINE] Available at: http://codact.ch/gb/cyclogb.html. [Accessed 25 April 2015].
- Berselli, G., Piccinini, M. and Vassura, G. (2011). Comparative Evaluation of Selective Compliance in Elastic Joints for Robotic Structure. IEEE International Conference on Robotics and Automation. p4626-4631.
- Leatham-Jones, B. (1987). Industrial Robot Configuration. Elements of Industrial Robotics. London: Pitman. p23-50.