Robots Cooperative Control with Disturbance Observer - Research Paper Example

 From the first serious exploration of cooperative robotic systems in the 1980s, these systems have rapidly evolved from the first dual axes robot arms to the complex and highly specified robotics used in industrial settings today (Liu and Wu 2001). Industrial robots are generally reprogrammable, multipurpose machines with some ability to sense, plan, and act to complete defined functions, and they generally have a contained mechanism for observing and addressing disturbance. These robotics can often carry out motion in three axes and utilize a broad range of manipulators. They comprise one of the largest and most diversified classes of machines that are able to autonomously interact in an industrial environment. Additional challenges arise when programming for complex or specific processes between multiple robotic systems, for which many solutions exist modeled after both high-level human-interactive robotics and biological systems. Observing and controlling disturbance in this varied group of robots provides a unique challenge for engineers and researchers. Understanding cooperative control in current industrial robotics and in prototypes that may become part of robotics technology in the future is paramount to addressing disturbance in these systems. Though taxonomies in the field of cooperative robotics are rapidly changing, Dudek provides a convenient classification and metrics scheme for these systems by classifying robotic collectives by seven different characteristics: collective size, range of communication, communication topology, and communication bandwidth, collective reconfigurability, processing ability of each agent, and collective composition (Dudek et al. 2002). Most cooperative control systems are designed as either swarm-type or independently intelligent. That is, the agents in cooperative systems may be designed to operate with low level instructions intact or to operate with no low level instruction at all, transmitting instead to a base station for planning and organizational control (Caviggia et al. 2002). The two types of systems can be distinguished by application of Matarić’s definitions of explicit and implicit cooperation in which explicit cooperation occurs when on agent takes action to benefit another agents goal, whereas in implicit cooperation each agent acts selfishly to complete its own goal, which modifies the environment in ways beneficial to goal completion for other agents (Matarić 1995). Radio signals, succinct macro messages, and even physical signals such as gestures or contact have long been used in communication between robotic agents (Raghunathan 2006). Advances in wireless technology have resulted in simple and effective new ways of relaying information to a central base station for processing cooperative actions (Ray 2005). One of the biggest topics facing researchers and developers of cooperative robot systems for industry is developing new and effective modes of communication, which has also been explored by software simulation (Wilson 2009). Simulation-based study provides an opportunity to examine multiple interactions in cooperative systems and easily make modifications that allow researchers and developers to experiment, understand, and evaluate intelligent robotic systems before undertaking the timely and costly development processes necessary for application to industry (Hu and Zeigler 2004). Simulation also provides a viable means to test theories on hoe disturbance can be handles, resulting in unique approaches, such as digital disturbance compensation for high speed and underwater activities (Yatoh et al. 2007). In industry, offline programming of coordinated systems helps to maximize return on investment (Zhang and Heping 2010). Swarm Intelligence (SI), in which a large number or very simple individuals aggregate by self-organizing algorithms in order to produce complex coordinated behavior, is being investigated as a method for stimulation of the action of coordinated systems (Caviggia et al. 2002). Swarms often operate on the idea that a signal left in the environment by a previous action will stimulate the performance of the next action by another agent of the swarm, propagating behavior and leading to the emergence of apart systemic behavior, a type of bevior called stigmergic action (Parunak 2003). Though swarm research in the United States is primarily defense related, several European collaborative projects are now investigating the practicality of applying SI technology to autonomous, mobile multi-robot systems for industrial use (Bogue 2008). These robots are generally well suited to simple tasks like collecting rocks or sorting mail, and are currently of limited industrial use. Other systems, such as dynamically reconfigurable robotic system (DDRS), have also been developed to control cooperative systems by exploring how nature controls multiple element systems. These systems are based on biological cells, with a large number of very simple robotic “cells” having one fundamental mechanical function, such as gripping, joining, or providing support. Together they interact in order to complete complex activities, including the control of disturbance (Fukuda, T and Nakagawa 1987). Based on human vision but with the capacity to record, store and analyze ‘sights’ more effectively than any human operator, Machine Vision (MV) has become one of the most effective new technologies in allowing machines to both report, document, and inspect for failures but also to localize, communicate, and interact in cooperative systems (Davies 2005). Automated Visual Inspection (AVI) is commonly used in industrial settings to monitor quality control (Pham and Alcock 2003). Alternatively, cooperative systems that are based on internal intelligence generally contain a smaller number of agents that operate at more sophisticated level. These agents are aware of the actions of nearby agents and their local environment and adjust their actions accordingly (Dadios and Maravillas 2002). In these systems the agents may be either homogeneous or heterogeneous. Heterogeneous agents, where the individuals are all distinct, allows for more specialized tasks, though they require greater development costs and are more difficult to reprogram (Matarić et al. 1995). Most industrial cooperative agents are of the heterogeneous variety, as they have highly specialized purposes (Ray 2005). Simultaneous Localization and Mapping systems (SLAMs), which remain an extremely dynamic field in research, provides another possible solution for either swarm or independently intelligent cooperative robotics. These technologies allow a robot to complete actions by building a virtual map of the environment in which it is operating and identifying its own ego-position in the continuously modified map, allowing these robots to quickly and effectively be adapted to new and varied tasks (Boga 2010). These technologies have the ability to use probability based systems to adjust for disturbances based on their own interaction with their environment, and as such are simply reprogrammed and applied to a broad variety of uses, making them particularly interesting in industrial applications. A number of cooperative robotic technologies have proven particularly effective in industrial settings. ACTor-based Robots and Equipments Synthetic System (ACTRESS) is a distributed cooperative system of many robots used for maintenance in nuclear power plants, where radiation exposure may prevent human technicians from being practical. The autonomous components of this system can communicate with each other, evaluate their own condition, sense their environment, make decisions, and act on those decisions (Asama et al. 1989). Important research in cooperative control systems is also occurring in Multiple Autonomous Robot systems (MARs), where graph-based algorithms allow the individual elements of these systems to discover, accomplish cooperative localization, and complete complex tasks as a group—tasks such as search and rescue or even fire fighting. Omnidirectional cameras and other forms of visual sensors can be used to allow localization of these robots to occur by visualization and coordination to occur by selection of a leader based on predefined parameters (Caviggia et al. 2002). These technologies are applicable to dynamically centralized control systems often found in industry as well as non-global relative localization and controlled formation methods across robotics. Observing and controlling the disturbance of manipulators in cooperatively controlled robots is a unique challenge because the forces in three dimensional space may act independently on different agents in the system. Disturbances caused by friction, vibrations, torques in torsional systems, impulses caused by surface contours, and many other forces must be considered and corrected in order to apply robotics to many specialized industrial applications. Disturbance Observers (DOs) have been commonly used in robotics to detect unknown or uncertain disturbance without the use of additional sensors using a linearized or linear system techniques, but, in order to be effective in multirobot systems, nonlinear models have also been developed (Chen et al. 2000). In addition, a number of researchers are developing methods of reducing disturbance effects on the system by controlling agents, such as the method developed by Dr. Oh and his group in which only a minimal parametrization of null space is performed, which is then augmented by conventional velocity relationships to determine the extended task space formulation (Oh and Chung 1999). Industrial applications for DOs are varied. In many industrial settings, torsional systems are common, and DOs can be used as part of the control system to suppress vibrations in these systems (Hori et al. 1996). Many industrial applications call for the use of robotic systems with independent joints controlled by multilink manipulators, and many of these systems use DOs to adjust for disturbance, and many older systems could benefit by the addition of DOs (Chen et al. 2000). DOs can also be used to estimate and compensate for friction in observer-based control systems, which is one of the most financially promising applications of DO systems (Chen et al. 2000). Advanced DO systems have been designed to account for minute time delays and eliminate feed-forward gains in highly specialized industrial processes, such as semiconductor packaging, that requires accuracy to the micron level (Kempf et al. 1999). DOs can improve the performance of industrial robotics systems and numerous counts of unique applications of these systems appear in contemporary research. Control theory is an interdisciplinary branch of mathematics combined with electrical engineering that seeks to understand the behaviors of dynamical systems, such as cooperative robotics. The reference is the desire output of a system, and the controller generally manipulates the inputs over time to achieve the desired effect on the output of the system (Leigh 2004). As the real world is far from ideal, controlling coordinated systems is not a black and white issue. These robotics are physical systems, and as such as subject to the same laws of motion and other mechanical object and driven by continuous actuators. Motion planning and control of these systems requires formulating the motion planning problem as an unconstrained variational problem, often involving sophisticated optimal control and calculus variations (Desai 1998). Developing the general framework for these controls can be complex, especially for industrial robots designed for highly specialized tasks. Cooperatively controlled industrial robots, among others, have become capable of greater operating work volumes and better repeatability accuracy in the past decade, as well as dropping to a price range practical in many industrial settings. A typical industrial robot that cost $100,000 ten years ago, now costs $30,000, making it much more attractive to medium sized manufacturing and industrial companies (Colestock 2008). In addition, the quality and control of robotic manipulators has improved, allowing for complex movements as well as implementation of micro-scale and bench-top robots using Machine Vision processes, now found in scientific labs (Digital Inspection Systems 2005). Simple arm joints are still widely used, with joints that may be actuated by motors or hydraulic actuators, though may newer arm geometries and joints allow for easier maintenance and more detailed, reprogrammable motion. Improvements in robot control, planning, and human interfacing are necessary to allow cooperative robot systems to be a practical alternative as manufacturing continues to shift from the high volume assembly-line and conveyor belts of the mid-twentieth century, to the contemporary High-Mix Low-Volume (HMLV) manufacturing seen in current industrial environments. Advances in coordinated robotic systems control coupled with more capable manipulators have made these systems more practical for industry (Lewis et al. 2003). As robotics, particularly with the onset of development of cooperative robotics, have continued to evolve, robotic control systems have simultaneously evolved to accommodate the changes that industrial demand has brought to this technology (Khairallah 2007). A new challenge in cooperative robotic control is the development of concurrent synchronization techniques necessary for operating multiple groups of industrial cooperative robots simultaneously. Distributed and decentralized strategies have been developed that further extended to adaptive synchronization and partial-state coupling (Chung and Slotine 2009), which may be applied in the future to the development of complex systems for industry, such as cooperative systems of robots capable of providing maintenance to other robotic systems. Cooperative robotics is a multi-disciplinary field that spans the full range of computer science, electrical engineering, and artificial intelligence, biology, communication, and robotics, all with an emphasis on coordinated group control (Kumar et al. 2003). Continuing development of the manipulators and control systems that allow cooperative systems to be effective at controlling disturbance and being able to accomplish increasingly detailed and specific tasks is necessary for implementation of these systems in industry. Reference 1. Asama, H, Matsumoto, A, and Ishida Y 1989, 'Design of an autonomous and distributed robot system – ACTRESS', in Proceedings of the IEEE/RSJ International Workshop on Intelligent Robots and Systems, pp. 283-290. 2. Boga, S 2010, 'Vision-Enhanced Localization for Cooperative Robotics', Thesis Project, Auburn University. 3. Bogue, Robert 2008, "Swarm intelligence and robotics’, Industrial Robot: An International Journal, vol. 35, iss. 6, pp. 488 – 495. 4. ‘Benchtop robots work with machine vision’ 2005, A Digital Inspection Systems Product Story, viewed online at http://www.manufacturingtalk.com/news/dli/dli110.html 5. Caviggia, K, Deming, J, Memon, M, Milner, A, and Thomas, N 2002, ‘Cooperative Robotics’, Literature Survey Presented to the ISRG, viewed 8 August 2010. 6. Chen, W, Ballance, D, Gawthrop, P, and O'Reily, J, 2000, 'A Nonlinear Disturbance Observer for Robotic Manipulators', IEEE Transactions on Industrial Electronics, vol. 47, no.4, 932-938. 7. Chung, S, and Slotine J 2009, 'Cooperative Robot Control and Concurrent Synchronization of Lagrangian Systems', IEEE Transactions on Robotics. 8. Colestock, H 2008, Industrial Robotics, McGraw Hill Publishers, New York, NY. 9. Dadios, E P, and Maravillas O A 2002, ‘Cooperative mobile robots with obstacle and collision avoidance using fuzzy logic’ in Proceedings of IEEE International Symposium on Intelligent Control, pp 75-80. 10. Davies, E R 2004. Machine Vision: Theory, Algorithms, Practicalities. Morgan Kaufmann. 11. Desai, J 1998. Motion Planning and Control of Cooperative Robotic Systems. Institute for Research in Cognitive Science IRCS Technical Reports Series. University of Pennsylvania. 12. Dudek, G, Jenkin, M, and Milios, E 2002, ‘A Taxonomy of Multirobot Systems’, Robot Teams, Edited by Balch, T., and Parker L.E., A K Peters, viewed 8 August 2010. 13. Fukuda, T and Nakagawa, S 1987, ‘A dynamically reconfigurable robotic system’, in Proceedings of the International Conference on Industrial Electronics, Controls, and Instrumentation, vol. 2, pp. 588-595. 14. Hori, Y, Chun, Y, and Sawada, H, 1996, ‘Experimental Evaluation of Disturbance Observer-based Vibration Suppression and Disturbance Rejection in Torsional System’, viewed 11 August 2010. University of Tokyo, Japan, viewed 11 August 2010. http://mizugaki.iis.u-tokyo.ac.jp/old-papers/hori/PEMC96.pdf 15. Hu, X, and Zeigler, 2004. 'Measuring Cooperative Robotic Systems Using Simulation-Based Virtual Environment', NSF grant DMI- 0122227. 16. Khairallah, O H 2007, 'Robotics Control Using Active Disturbance Rejection Control', Thesis Project, Cleveland State University. 17. Kempf, C J, Kobayashi, S, and Richmond, CA, 1999, 'Disturbance observer and feedforward design for a high-speed direct-drive positioning table. Control Systems Technology', IEEE Transactions. vol. 7, iss. 5, 513-526. 18. Kumar, V, Leonard, N, and Morse, A, Stephen 2004. 'Cooperative Control A Post-Workshop Volume 2003' Block Island Workshop on Cooperative Control, vol. 309. 19. Leigh, J R 2004, Control theory, Institution of Electrical Engineers. 20. Lewis, F, Dawson, D, and Abdallah, C 2003, Robot Manipulator Control: Theory and Practice, CRC Press. 21. Liu, J, and Wu, J 2001. Multi-Agent Robotic Systems. CRC Press, London. 22. Matarić, M J 1995, ‘Issues and approaches in the design of collective autonomous agents’, Robotics and Autonomous Systems, vol. 16, pp 321-331. 23. Matarić, M J, Nilsson, M, and Simsarian K T 2005, ‘Cooperative multi-robot box-pushing’, Proceedings of IEEE/RSJ International Conference on Intelligent Robots and Systems, vol. 3, pp. 556-561. 24. Oh, Y, and Chung, W K, 1999, 'Disturbance-observer-based motion control of redundant manipulators using inertially decoupled dynamics', Mechatronics, vol.4, iss. 2, 133-146. 25. Pham, DT and Alcock, RJ 2003. Smart Inspection Systems: Techniques and Applications of Intelligent Vision. Academic Press. 26. Parunak, H. v D. 2003, 'Making swarming happen', In Proc. of Conf. on Swarming and Network Enabled Command, Control, Communications, Computers, Intelligence, Surveillance and Reconnaissance (C4ISR), McLean, Virginia, USA. 27. Raghunathan, Arun 2006. 'Ad-Hoc And Multi-Hop Wireless Sensor Networks for Activity Capture in Cooperative Robotics', Thesis Project, Auburn University. 28. Ray, A 2005, 'Cooperative Robotics Using Wireless Communication', Thesis Project, Auburn University. 29. Wilson, C 2009. 'Hardware Testbed for Collaborative Robotics using Wireless Communication', Thesis Project, Auburn University. 30. Yatoh, T, Sagara, S, and Tamura, M 2007, 'Digital type disturbance compensation control of a floating underwater robot with 2 link manipulator', Artificial Life and Robotics, vol. 13, iss. 1, 377-381. 31. Zhang, B, Heping, C 2010, ‘Industrial Robotics and Automation Workshops: CASE 2010’, 6th IEEE Conference on Automation Science and Engineering, Toronto, Canada, viewed 8 August 2010. Read More