Servo-pilot for autonomous driving

• Autor(a) principal: Ávila, Emanuel da Silva
• Data de Publicação: 2010
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10773/2537
• topic: Autonomous driving; Micro-electromechanical systems (MEMS); Modularity; Inertial systems

This project is part of a broader activity that aims to develop a setup to allow a car to have perception about the driver’s actions on the vehicle’s controls and latter to control the vehicle in navigation situations. The first part of the project will allow monitoring the vehicle controls, such as pedals, and the setup will be flexible enough to be installed and removed easily. Also, it will not require vehicle modifications and can be used for other applications that require similar types of measure- ments, such as antropometry during electro-stimulation therapy or to monitor humanoid robot motion and balance. The technological choice has been directed to inertial-based sensors such as accelerometers and gyroscopes due to their attractive cost, small packaging size and absence of moving parts, which improves durability. One novelty on the proposed setup is that many inertial sensors can be connected in the same network and differential measurements can be done to decouple the local motion of parts from the vehicle global dynamics. Also, the size of the devices developed is much smaller and lighter than all com- mercial solutions found. For this purpose, the project relies on MEMS accelerometer and gyroscope modules connected in a master multi- slave network using CANbus to support the communications for motion analysis. On the other hand, the project includes an industrial PLC connected to a combination of pneumatic cylinders and electric pres- sure controllers to emulate the human foot actuation on the car control pedals. The network of inertial sensors proved to work well on near- stationary situations where the dynamic acceleration applied to all the modules is negligible. When subject to higher dynamic perturbations, the system requires further software improvements so it can perform adequately.