Data driven time to collision model for unmanned aerial vehicle control system under various payload and speed conditions
Time-to-collision (TTC) can be defined as the time required for vehicles to collide with another vehicle or static obstacle if they continue at their present speed and on the same path. Hence, the mathematical model of TTC is useful to assist the collision avoidance system (CAS) in any type of auton...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2023
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/102779/1/SulaimanSabikanPSKE2023.pdf.pdf http://eprints.utm.my/id/eprint/102779/ http://dms.library.utm.my:8080/vital/access/manager/Repository/vital:152262 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Time-to-collision (TTC) can be defined as the time required for vehicles to collide with another vehicle or static obstacle if they continue at their present speed and on the same path. Hence, the mathematical model of TTC is useful to assist the collision avoidance system (CAS) in any type of autonomous vehicle. This thesis, presents the data-driven TTC model for unmanned aerial vehicles (UAV) control systems under various payloads and speeds condition. The research consists of three phases. The first phase involved the design and development of a data logging system in the multirotor UAV platform. The data acquisition process for model development requires a UAV system, which consists of the quadrotor vehicle structure, onboard flight mission controller and a ground control system. The open sources platform UAV system development and Proportional–Integral–Derivative (PID) controller used for position, altitude and attitude control have been implemented. Experiments are conducted to collect the required flight data in an uncontrolled environment using a developed platform that has been recognized for its performance. In the second phase involved modelling TTC. Controller time stamps, radio control signal magnitude, global positioning system platform and speed parameters are recorded from different payloads, ranging from 0g to 200g. A data filtering algorithm was applied to eliminate data that does not meet the specified minimum horizontal speed. Particles Swarm Optimization (PSO) algorithm was used for optimizing the model and validating with the real data from the experiment. The collected onboard real experimental data for five different payloads have been analysed to develop a mathematical model of TTC through the PSO approach. Based on the experimental data, the fitness function relationship is considered to solve optimization between speed (m/s), payload (g) and their time-to-collision (s). The TTC model predicts the time required for the collision with a static obstacle based on its current flight parameters, such as speed and payload. Finally, the third phase involved the evaluation of the UAV control system with the TTC model throughout the simulation. The TTC model has been implemented in the UAV’s PID controller. Parameters such as initial speed, activation obstacle distances and final distance are introduced in the discussion of this thesis. Based on the workspace simulation environment that has been designed, the TTC model is applied to show the proposed speed based on the UAV's current speed. The activation obstacle distance obtained is a minimum of 5 metres with an initial speed of 2.0 m/s and the proposed speed will be given by the model, continuously. The distance between the obstacle and the reaching point is influenced by the payload. The distance without load is 2.589 metres, and the distance with a 200g load is 1.989 metres, both of which are safer than the specified final distance of 1 metre before a collision. In conclusion, the proposed TTC model has successfully determined the optimal proposed speed based on their current flight parameters under various payload and speed hence, it can be used as a risk assessment metric in UAV’s CAS. |
|---|