Real-time visual-inertial simultaneous localization and mapping (SLAM) for multirotors | |
| Author | Vilasinee Rasmeehirun |
| Call Number | AIT Thesis no.CS-18-04 |
| Subject(s) | Airplanes--Control systems Sensor networks Robots--Control systems |
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Computer Science, School of Engineering and Technology |
| Publisher | Asian Institute of Technology |
| Series Statement | Thesis ; no. CS-18-04 |
| Abstract | This thesis is an implementation of real-time VIORB SLAM on board for multirotors. The system includes two main parts which are the web application as a ground station application and onboard application as an autonomous control and VIORB SLAM system. This system tries to achieve low-cost multirotors platform for 3d reconstruction using visual inertial SLAM (VIORB SLAM) with the abilities of Pixhawk flight controller and PX4 firmware for IMU data and navigation. The ground station web application is a part of navigating the multirotors in study areas as same as waypoint navigation and aircraft management system including path and part management. The communication between the ground station and aircraft is done over the internet connection or WiFi connection by SSH provide abilities to control and display stage of the aircraft real-time in a long distance than radio communication. The onboard application is divided into two parts. First, an autonomous control system using MAVLink take part of generating, sent and receive MAVLink messages with additional logic to check the state of the aircraft, current positions, takeoff and landing status. Second, VIORB SLAM system responsible for real-time 3d reconstruction. Both parts simultaneously cooperate with location and camera manager, feeding IMUs and image data to SLAM for real-time 3d reconstruction and recorder for dataset collection. Time synchronization is a big issue for working with multiple sensors. The visual-inertial SLAM, require IMU and image as input, the system solves this problem by using Unix time as a standard timestamp format of the system and finding time offset delay from camera to IMU. SLAM system can work in both indoor and outdoor environment such as room, street, and garage, but not perfectly work under tree farm environment depend on kind of plants. The dynamic scene directly affects the estimation process led to wrong trajectory path estimation. |
| Year | 2018 |
| Corresponding Series Added Entry | Asian Institute of Technology. Thesis : no. CS-18-04 |
| Type | Thesis |
| School | School of Engineering and Technology (SET) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Computer Science (CS) |
| Chairperson(s) | Dailey, Matthew N.; |
| Examination Committee(s) | Soni, Peeyush;Mongkol Ekpanyapong; |
| Scholarship Donor(s) | Royal Thai Government Fellowship; |
| Degree | Thesis (M. Eng.) - Asian Institute of Technology, 2018 |