In the development of robots to intelligence, multi-robot collaboration system is a kind of coverage technology integration platform. If the intelligence of a single robot is only to make the individual person smarter, then the multi-robot collaboration system not only has a group of smart people, but also requires them to cooperate effectively. Therefore, it not only reflects individual intelligence, but also reflects collective intelligence. It is an imaging and innovative exploration of human society's production activities.
The multi-robot collaboration system has a wide range of applications. It is closely related to the expansion of automation to non-manufacturing. As the application environment shifts to unstructured, multi-mobile robot systems should be able to adapt to task changes and environmental uncertainties. With a high degree of decision-making intelligence, the research on multi-mobile robot collaboration is not simply the coordination of control, but the coordination and cooperation of the whole system. Here, the organization and control of the multi-robot system largely determines the effectiveness of the system.
Multi-robot collaboration systems are also a model for implementing distributed artificial intelligence. The core of distributed artificial intelligence is to divide the whole system into several intelligent and autonomous subsystems. They are physically and geographically dispersed, and can perform tasks independently. At the same time, they can exchange information and coordinate with each other through communication, thus completing the overall task. This is undoubtedly attractive for the completion of large-scale and complex tasks, and soon gained widespread attention in military, letter and other applications. Multi-machine collaboration system is the concrete realization of this concept, in which each robot can be regarded as an autonomous agent. This multi-smart robot system MARS (MulTI-AgentRoboTIcSystems) has become a new research hotspot in robotics. .
The multi-mobile robot system has the mobile function and can complete complex tasks in the non-structural environment. It is the most typical and application prospect in the multi-robot collaboration system, and it is also the most widely studied system. The following is a representative of the key technologies of the intelligent robot collaboration system, represented by multiple mobile robot systems:
1. Architecture
The architecture is the logical and physical information relationship and control relationship between the robots in the system, and the distribution model of the problem solving ability. It is the basis of the cooperative behavior of multiple mobile robots. Generally, the architecture of a multi-mobile robot collaboration system is divided into two types: centralized (distributed) and distributed (distributed). The centralized architecture can be planned with a single master robot with all the information about the system's activities. The distributed architecture does not have such a robot, in which all robots are equal with respect to control. Although the centralized architecture can achieve global optimal solution, in fact, people prefer the distributed structure because of the uncertainty. In recent years, in the distributed architecture, in order to overcome the difficulty of modeling the environment in the actual environment of the robot, and to improve the robustness and operation ability of the multi-mobile robot cooperation system, some scholars have adopted behavior-based reactive control of physical strength. Some scholars adopt a behavior-based reactive control architecture, which establishes a cooperative behavior in a reaction mode, accelerates the response of mobile robots to the outside world, avoids complex reasoning, and improves the real-time performance of the system.
2. Perception
Perception is the basis of intelligent robotic action, including “feeling†(sensing) and “knowing and understanding†information fusion and utilization. The most important perceptual problem in mobile robots is the positioning and environment modeling problem [7]. Although there are odometer estimation, vision-based road sign recognition, map-based global positioning, gyro navigation, GPS and other positioning methods, However, in an unknown non-structural environment, GPS is currently available for practical global positioning. However, GPS is limited by factors such as accuracy and safety. How to improve positioning and environment modeling capabilities by means of cooperation between robots is an important part of researching the intelligence of multi-mobile robot systems. In recent years, a variety of simultaneous methods for establishing and locating environmental maps have been proposed [8], in which the environment modeling and positioning process are accompanied by each other, and the two are gradually clear in the process of iterating each other, but often require harsh environments. condition. In addition, in many collaborative tasks, only relative position information between partners, such as formation and local collision avoidance, is required. Therefore, local positioning based on sensors is also concerned. Robots use ultrasonic, infrared, laser or visual sensors to interact with each other. The detection, and then through the statistics, filtering and other algorithms for information fusion, thereby obtaining the relative position of each robot in the system.
3. planning
The planning problems mainly include task planning and path planning. It has been the main problem of artificial intelligence and robotics research. It has carried out extensive and long-term research, and the results have been applied in the research of planning problems of multi-robot collaborative systems. Corresponding to the structure, the planning of multi-mobile robot systems usually includes two methods: centralized planning (Centra)-ized1anning (DistribUted “nningâ€). Centralized planning generally achieves efficient and globally optimal planning. The result, but it is mainly suitable for static environments, and it is difficult to cope with changes in the environment. In distributed planning, each robot plans its own actions according to the environmental information it owns. Its advantage is that it can adapt to changes in the environment. Excellent solution and possible deadlock.
4. Learning and evolution
Learning and evolution are the embodiment of the system's adaptability and flexibility. At present, the collaborative learning (Reinforcement Learning) method and genetic programming (GeoeTIcPrgramming) are mainly used in collaborative robotics, and have been successfully applied in multi-robot handling systems and robot soccer [10] [11]. The current multi-robot learning and evolution still stays at a relatively low level of behavior. The tasks and environments for learning and evolution are also very simple. When faced with more complex tasks and environments, there are time-delay evaluations and combinatorial explosion problems. In addition, the distributed learning and evolution of multi-agents is also significantly different from the traditional centralized learning and evolution methods. It is still necessary to find more effective behavior optimization methods.
5. Coordination and collaboration strategy
Multi-mobile robot systems involve the coordination of robot tasks, planning, and control when collaborating to complete complex tasks [12][13]. The research on multi-agent theory has provided ideas and strategies for these coordinated behaviors, but how to These abstract ideas and strategies are combined into specific systems to achieve, while at the same time reflecting universality, and what tools are used to correctly describe system behavior at all levels. At present, the most typical description tool in the task coordination layer is the finite state machine (FSA) method in the discrete event dynamic system theory. However, how to describe the mixed system theory and method for different levels of behavior is still a hot topic in research. In addition, multiple mobile robots operating in the same environment often have conflicts in resource utilization. If there is no proper coordination strategy, the system will not work properly. For foreseeable conflicts, it can be avoided by planning. However, the dynamic operation of the system is often not accurately predicted in advance, and the conflict resolution will be limited only by means of planning. The resolution of dynamic conflicts mainly includes consultation method, custom method (ConvenTIon) and acquaintance model method. Deadlock detection and resolution in dynamic environments is still a very challenging problem.
6. System software platform development
The research on multi-robot systems has been going on for nearly 20 years. The previous work mainly focused on the research of system hardware and some related single technologies. With the gradual improvement of multi-mobile robot hardware systems, the current software research is obviously lagging behind. The developed software is often targeted at specific hardware systems and single tasks. The technology integration is low and the versatility is poor, and the hardware performance cannot be effectively utilized. To this end, people urgently need to develop a system software platform with high openness, versatility, robot hardware independence and scalability, system integration of existing scattered technology results, and standard for designing system software design framework. . In the past three years, the United States and European countries have launched a number of large-scale projects for software development of multi-mobile robot collaboration systems, and have produced some representative software development platforms that have been applied.
7. Experimental Study
The experimental research of multi-mobile robot collaboration system started from the simulation of computer and used computer software to build a hypothetical robot group. In this way, the robot body can be freely given the ideal mechanism to interact in different ways. However, although this method can examine the influence of many rational or biological principles on the cooperative behavioral norms of robot groups, it is difficult to directly apply to constructing actual operating systems. In recent years, with the improvement of the performance of robots and their components, the research on the use of real-machine multi-robot systems has been increasing, so that the distance between theoretical research and the actual environment and existing physical robots has been gradually reduced. At present, most of the international robotic basic research laboratories are conducting research from both computer simulation and real machine experiments.
The Cat6a doubles data transmission bandwidth, from 250 to 550 MHz; decreases the chance of crosstalk interference; and provides superior reliability and transmission speeds through greater lengths of cable.
A Category 6a Ethernet patch cable is also referred to as a Cat6a Network Cable, Cat6a cable, Cat6a Ethernet Cable, or Cat 6a data/Lan Cable. Future-proof your network for 10-Gigabit Ethernet (backwards compatible with any existing Fast Ethernet and Gigabit Ethernet); Meets or exceeds Category 6a performance in compliance with the TIA/EIA 568B.2 standard.
High Performance Cat6a, Shielded Ethernet Cable provides universal connectivity for LAN network components.
Product Information:
1. Frequency - 550 MHz
2. Transmission Speed - 10 GB
3. Available Length - 0.5M~30M or Longer
4. Connectors - Nylon Gold Plated RJ45 or Nickel Plated RJ45
5. Condcutor- 26AWG or 28AWG (America wire gague)
6. Cable Construction - Shielded or Unshielded
7. Jacket- LSZH or PVC
8. Install - Used indoor, in-wall and in the ceiling
Applications: Computer, Router, TV, Interchanger, Concentrator, ADSL, Set-top box And So on.
Cat6A Ethernet Cable,Cat6A Shielded Connectors,Cat6A Rj45 Connectors,Cat6A Patch Cables
Shenzhen Kingwire Electronics Co., Ltd. , https://www.kingwires.com