Task Programming of Redundant Industrial Robots: A Virtually Extended Null Space Formulation Verified Through Obstacle Avoidance
MetadataShow full item record
Industrial robots are an important part of modern automation and are used in a variety of applications such as handling, welding, painting and assembling. They have normally six degrees of freedom to provide an arbitrary location of the tool inside its working space. However, during the last years, there have been developed industrial robots with more than six axes, thus redundant robots. This gives extra functionality to avoid singular positions, to move around obstacles and to optimize the use of energy during a predefined trajectory, by changing the internal configuration of the robot arm while still maintaining the tool's location. However, programming of redundant robots is complicated and time consuming due to the fact that both the tool location and the internal configuration of the robot arm have to be programmed. Investigations on the use of redundant industrial robots in the industry reveal several advantages including highly increased flexibility and a significant reduction of space need. The flexibility can be attributed through obstacle avoidance, singularity avoidance and energy optimization. For small and medium enterprises (SMEs) and in High-mix Low-volume productions, this gives a great advantage in even more competitive markets. This thesis presents an efficient approach for programming of redundant industrial robots. The system uses proximity sensors mounted on the robot arm to detect obstacles. By analyzing the sensor data, the system can automatically reconfigure the robot's arm to automatically, comply with environmental constraints. Enabling this functionality by an automatic system, simplified the programming of the redundant robot to be similar to a normal six-axes robot. Studies into the subjects that constitute the theoretical basis for the practical implementation and kinematic resolutions have been done. This includes, among others, kinematic analysis, redundancy resolutions, task formulation and methods for obstacle detection. The studies led to a suggestion for an task description scheme based on an extension of Mason's task formulation for force controlled tasks. The formulation augments the robots self-motion ability to be based upon a virtual extension of the robot's Null space. The virtual extension allows the operator to select the priority of the secondary task, subsequently programming the robot as if it were a six-axis robot. The system has been implemented and experimentally verified on a NACHI MR20 seven-axes industrial robot. The implemented system includes Cartesian velocity limiter, Joint space velocity limiter, Task Reconstruction algorithm, Default arm reconfiguration and path correction algorithm. The sensors system is based on ultrasonic and infrared proximity sensors, covering the greater part of the robot arm. The experiments proved convincing performance and robustness of the implemented system. It was shown that the extended null space formulation can redistribute certain axes from the primary task to the secondary task, and thus, provide automatic obstacle avoidance. The obstacle avoidance strategy was shown to be successful, and gave the desired evasive maneuver. Experiments also demonstrated the system’s ability to reconfigure the primary task after deflection caused by the secondary task and the ability to reconfigure the arm to a default configuration when both the task is reconstructed and no obstacles are present.