| Summary: | 碩士 === 國立成功大學 === 資訊工程學系 === 103 === The main purpose of the research is to allocate the balls controlled by motors into a pre-designed dynamic pattern and to interact with human. In order to reach this goal, several attempts were made to build the system, which includes an introduction of a man-machine interaction system and an improved motor control accuracy.
Typical problems with such interactive systems for multiple motors include inaccuracy of motors, delay in transmission, inaccuracy of motion detection and even a prolonged response time of motors. These problems may contribute to an unsatisfactory user experience or an unstable pattern demonstration. Therefore, a modified interactive system for multiple motors was proposed. The system was designed in a way that makes it able to demonstrate a dynamic pattern-running system and interact with human (satisfactory user experience expected) under same structure and the system must be capable of extending easily. To reach such goals, several improvements were made in the thesis:
1. Initialization design that enables operators to increase the number of slave boards easily
2. An algorithm to predict users’ joint movements, allowing the motor to adjust accordingly
Otherwise, the code of the system was written in a way that minimizes the future need for modification should a change to the hardware takes place.
The system adopts 32 stepper motors, 3 Microchip® PIC18F4580 demo boards, 8 NXP Semiconductors® 74HC595 shift registers, 32 Allegro MicroSystems® A4988 microstepping driver and a Microsoft® Kinect. In interaction mode, Kinect records users’ joint movements and send the results to the master board. These data will be analyzed and send to slave boards. Each board is connected by CAN bus. The communications between PIC18s and 74HC959s are through SPI. In pattern-running mode, the patterns are stored in the master board or PC and the periodically recorded data are dispatched to slave boards.
The experiment results are shown in algorithm evaluation, which is calculated based on the score given by participants after double-blind tests.
|
|---|
