Techniques For Sensor-Integrated Robotic Systems: Raman Spectra Analysis, Image Guidance, And Kinematic Calibration

摘要

Robotics and sensor technology have made impressive advancements over the years. There are now robotic systems that help perform surgeries or explore the surface of Mars, and there are sensors that detect trace amounts of explosives or identify diseased human tissue. The most powerful systems integrate robots and sensors, which are natural complements to each other. Sensors can provide information that might otherwise be unavailable due to indirect robotic manipulation (e.g., images of the target environment), and robots can provide suitably precise positioning of an analytical sensor.To have an effective sensor-integrated robotic system, multiple capabilities are needed in the areas of sensors, robotics, and techniques for robot/sensor integration. However, for many types of applications, there are shortcomings in the current technologies employed to provide these capabilities. For the analysis of complex sensor signals, there is a need for improved algorithms and open platforms that enable techniques to be shared. For the path planning and tracking of integrated sensors and the visualization of collected information, image guidance systems that support advanced analytical sensors would be very beneficial. For robotic placement of a sensor, easily usable calibration procedures and methods to overcome limited feedback could help improve the accuracy.To help address these issues, some novel systems and techniques were developed in this research. First, a software system was created to process, analyze, and classify data from a specific kind of sensor (a Raman spectrometer). The system is open and extensible, and it contains novel techniques for processing and analyzing the sensor data. Second, an image guidance system was made for use with a sensor-integrated robotic system (a Raman probe attached to a surgical system). The system supports tool tracking, sensor activation, real-time sensor data analysis, and presentation of the results in a 3D computer visualization of the environment. Third, a kinematic calibration technique was developed for serial manipulators. It requires no external measurement devices for calibration, provides solutions for some limitations of existing techniques, and can significantly enhance the positional accuracy of a robot to improve sensor placement.The implemented techniques and systems were successfully evaluated using various data sets and conditions. Together, the contributions of this work provide important building blocks for an accurate robot with an integrated analytical sensor. This type of a system would be a powerful tool for many future applications, such as a surgical robot that automatically scans for diseased tissue and assists the surgeon in the necessary treatment. Ultimately, this work is intended to foster the development of advanced sensor-integrated robotic systems.