A new eco-concept, combining mechanical design and biomimetic processing of measurements for control, providing new possibilities for collaborative robots and new perspectives to the field of cobotics in regards of the safety of physical human-machine interaction. This new technology will bring new robotics solutions to the problems of medical isolation, maintaining dependents at home, and support on arduous industrial tasks.
The major objectives of the sectors of manufacturing and handling services concern the improvements in terms of reducing financial costs, increase quality and productivity; new societal and environmental issues should be taken into account: safety and environmental impact and lack of skilled labor for the production of heavy or highly specialized tasks. A possible way to meet these challenges consists in the use of collaborative robots (cobots) to assist and work with human operators. But proximity between robots and people raises problems related to the safe operation of robots. These brakes could be removed through appropriate technologies at low cost that make market robots inherently safe. The solution proposed in this project is a generic mechatronic device with its innovative operating systems performing the function of biomimetic joint. This system will allow cobots to adapt their behavior to secure the environment and the task by ensuring the adaptive damping functions in case of impacts like a biological articulation, modeling of physical contact and fast data transmission. The scientific objectives are to provide a compliant and compact device to adaptively manage the impacts on segments of a robot, to estimate online the mechanical contact impedance for a predictive control managing security. This will consist in selecting from the measurements, the best model structure and afterwards identifying the parameters. Physiological phenomena will be considered. Technological locks are related to the miniaturization of the product by suitable choice of mechanisms and materials, the search for compromise between computing power, space and thermal dissipation of embedded electronics.
A review of research projects on cobotics, as well as standards related to safety requirements for robots and measuring devices, allowed us to highlight the useful quantities and instructions for the definition of user needs and the SIS safety sensor. Following a literature review, the specification was completed and two types of torque response curves have been identified to meet the required safety function. For each type of response, a kinematic linkage has been proposed. The first compliant device, uses a six-bar mechanism and a linear behavior tension spring. It was developed in first version to be integrated on a sliding joint. Dimensional synthesis has led to the identification of the optimal design parameters allowing the mechanism to reproduce the desired torque curve. The second device uses a cam mechanism coupled to a reducer and tension springs. The proposed innovative solution provides a compact compliant device to be integrated on a revolute joint. An adjustable parameter thanks to an actuator provides multiple response curves characterized by threshold values of the input torque. Experimentations and analysis have been performed for the identification of mechanical parameters for the optimal design of the SIS. The experimental protocol has been designed to reveal key mechanical factors relevant in safe natural physical interactions between humans.
To verify that the experimental conditions of the chosen protocol correspond to «comfortable« situations for humans, we have included in the protocol the measurement of physiological parameters (heart rate, respiratory volumes) by means of the METAMAX system. The gestures analysis was performed with the motion capture system Vicon-Nexus.
A feasibility prototype of the first device was made that complies with an imposed compliant behavior model response, thus validating the methodology. Studies of this system continues by adapting it on the revolute joint and improving its integration on a robotic arm. A first design of the second device, and a feasibility model produced by additive manufacturing allows to validate the concept. The design that followed is based on a study of resistance and choice of materials on the ABAQUS software while taking into account the integration of the device on the robotic arm (SRA) that will be designed based on modules developed by PPRIME under EQUIPEX ROBOTEX.
The experimental works for the identification of mechanical parameters for the optimal design of the SIS, have provided results on limit values of deflection, velocity, force and power related to two distinctive functions of the SIS: i) robustness and performance, so that the SIS is able to sustain the nominal mechanical interaction parameters for long duration; ii) safety, so that the SIS enforces the respect of mechanical limits for the safety of persons.
A simulation platform was developed to study the overall behavior of a multi-articulated robot equipped with SIS devices. The models and parameters of the SIS are adjustable. The work is conducted on a SCARA robot for its didactic aspect. It will be extended to the case of a 7 d.o.f robot for medical application. A control with an adaptive admittance was developed and validated on the arm Kuka LWR with Shadow hand. This control automatically adjust the impedance parameters, to meet the ISO10218 standard for industrial applications (Maximum thresholds of speed, power, and strength at the «tool control point«). The behavior of the SIS was simulated on the joints of the robot Kuka LWR.
The Advanced CAD models to T0 + 18 allows to consider the achievement of one or two laboratory prototypes in the coming months. The first will validate the behavior and fix the choice of device. We will size the future prototypes based on i) their location on the robot and ii) on their target applications. The control electronics of SIS and its operating systems will be developed. The control with adaptive admittance will be extended to control the Shadow robotic hand and arm Kuka LWR. The gravity will be included in the simulation of the SIS in the joints, its proposed compensation will be under the dual mechanical / control approach. This will allow realization of a prototype for validation and a set of pre-industrial version of SIS. This set of sensors will be tested on a modular robot (SRA) for three applications covering the main areas of robotics, i.e. industrial robotics, medical robotics and assistive robotics. An evaluation report of SIS installed on the SRA will be provided at the end of the project.
1. F. COURREGES, J. ABSI, M. A. LARIBI, M. ARSICAULT, S. ZEGHLOUL «Accounting for Respiratory Motions in Online Mechanical Impedance Estimation», 2015 IEEE 13th International Conference on Industrial Informatics (INDIN), 22-24 July 2015, Pages 1504 – 1509, DOI: 10.1109/INDIN.2015.7281956.
2. F. COURREGES, M. A. LARIBI, M. ARSICAULT, S. ZEGHLOUL «An in Vivo Experiment to Assess the Validity of the Log Linearized Hunt-Crossley Model for Contacts of Robots with the Human Abdomen» The Fourth IFTOMM International Symposium on Robotics and Mechatronics held in Poitiers, France 23-24 June 2015, Pages 209-218, DOI: 10.1007/978-3-319-22368-1.
3. B. NAVARRO, P. KUMAR, A. FONTE, P. FRAISSE, G. POISSON and al. « Active calibration of tactile sensors mounted on a robotic hand ». IROS 2015: Intelligent Robots and Systems, IEEE/RSJ, Work-shop on Multimodal sensor-based robot control for HRI and soft manipulation, Hambourg, Germany, September 2015.
4. B. NAVARRO, A. CHERUBINI, A. FONTE, R. PASSAMA, G. POISSON, P. FRAISSE. « An ISO10218-compliant adaptive damping controller for safe Physical Human-Robot Interaction », IEEE International Conference on Robotics and Automation, ICRA 2016, Stockholm, Sweden, May 2016.
SISCob aims at developing a new intelligent and modular device mimicking the functions of biological articulations and their synergy for collaborative robots (cobots). This Safety Intelligent Sensor (SIS) device will connect the output shaft of an actuator to a robot link or to the end-effector when the actuator is in its terminal position; it will ensure the safe functioning of the robot itself and the safe interaction with the surrounding environment by providing the following services:
• an adaptive tunable compliant mechanical structure managing high frequency physical interactions;
• an online estimation of impedance models exploitable for predictive control;
• a real-time data transmission, allowing SISs to communicate as agents within an ad hoc body area sensor network that a central control unit can interrogate.
This project stems from the fact that manufacturing and manual services sectors rely on process improvements for reducing financial costs, raising quality and productivity. However, new societal and environmental constraints should also be accounted for: that is safety and environmental impact management and/or the lack of skilled manpower for specialized or arduous or hazardous work. To address these challenges, numerous R&D projects have focused on the usage of cobots to collaborate with and assist human operators. However, safety issues have slowed down the introduction of robots working in close vicinity with humans. These barriers could be broken down with suitable technologies allowing for safe physical man/machine interaction without hindering the versatility and engineering possibilities of robots. In contrast to the state of the art of compliance based techniques for robot safety, our aim is to achieve this level of safety independently of the robotic structure, actuation technologies and power along with the capability of robustly managing environmental uncertainties. These conditions are crucial for a real large-scale deployment on the market. We believe that the design of a very unique sensor could be the keystone of the solution. From a scientific point of view, two of the most critical identified issues are:
1. To propose a novel safety compliant articulated mechanism adapting its stiffness to impact forces on the robot links. This mechanism should be passive with the possibility to modulate actively its stiffness.
2. For a high quality robust control of the robot, the sensor should provide on-line a full model of the mechanical impedance the actuator’s shaft is interacting with. We will develop an innovative estimation approach, resilient to persistent excitation and consisting in two steps: a) Using an algebraic approach to select on-line the best candidate model from available data and b) Identifying the model’s parameters. Furthermore, a respiratory model could be combined with the impedance model to enhance its predictive capability for some specific applications.
Some advantages of the SIS include:
• Robot designers will have the possibility to conceive dependable and intrinsically safe robots with the SIS as a new low cost building block and without worrying about consequences on the actuators choice.
• The impedance model, estimated in real-time, will enable to supplement the safety functions by allowing the implementation of a fast predictive impedance control loop (reflex loop) of the actuators.
• When a robot is equipped with several SIS, the sensors network will permit to detect the impact/contact loci on the robot structure and trigger an appropriate global safety control strategy.
• Thanks to the delivered impedance model, the SIS will enable to use the efficient “model mediated teleoperation” technique in a new way for haptic teleoperation.
The "SIS" will be developed with a number of pre-industrial versions and tested on a modular robot for three applications covering three main robotics fields namely industrial robotics, medical robotics and assistive robotics.
Monsieur Marc ARSICAULT (PPRIME - Université de Poitiers)
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
PRISME PRISME - Université d'Orléans
UM2-LIRMM Laboratoire d'Informatique, de Robotique et de Microélectronique de Montpellier
PPRIME PPRIME - Université de Poitiers
XLIM XLIM - Université de Limoges
Help of the ANR 562,435 euros
Beginning and duration of the scientific project: September 2014 - 42 Months