Search

Obstacle detection and avoidance is desirable for UAVs especially lightweight micro aerial vehicles and is challenging problem since it has payload constraints, therefore only limited sensor can be attached the vehicle. Usually the sensors incorporated in the system is either type vision based (monocular or stereo camera) or Laser based. However, each of the sensor has its own advantage and disadvantage, thus we built the obstacle detection and avoidance system based multi sensor (monocular sensor and LIDAR) integration. On top of that, we also combine SURF algorithm with Harris corner detector to determine the approximate size of the obstacles. In the initial experiment conducted, we successfully detect and determine the size of the obstacles with 3 different obstacles. The differences of length between real obstacles and our algorithm are considered acceptable which is about -0.4 to 3.6.

Unmanned Aerial Vehicles (UAV) have a great potential to replace manned aircraft in many tasks or missions. However, in order to allow these UAVs to execute their mission appropriately and especially to deploy them in national or civil airspace, UAVs must have a proper and reliable collision avoidance system. This avoidance system is known as a Sense and avoidance (SAA) system. The purpose of the system is to detect any obstacles either in static or moving condition and respond with proper avoidance maneuvers in order to maintain minimum separation distances. Therefore the research project deals with the development of a unique SAA system which is based on multisensory integration. In particular, it focuses on sensors, processing logic and hardware required on the UAVs to acquire situational awareness. Dissimilar kind of sensors which is optical and laser type is proposed to be employed, so that weaknesses from each individual sensor can be tolerated. Proper logics or algorithm related to the fusion of output from each sensor will also be studied.The combination of Extended KalmanFilter (EKF) and Differential Evolution(DE) method is used for the purpose of development of detection and tracking system which considered to be the main challenges in SAA system development. Subsequently, once the algorithms are put together, including maneuvers avoidance technique algorithm, the developed SAA system will be experimentally tested in order to validate and evaluate the overall capability of the system.