Abstract
Pedestrian trajectory prediction is an important task in a range of applications such as social robots and self-driving vehicles. Although artificial intelligence research has made significant advances in the last decade, challenges still remain in automatically predicting the trajectories of pedestrians. This thesis explores different deep learning methods for the prediction of pedestrian trajectories. Novel data-driven methods are proposed to address key challenges including the social influence from other pedestrians, the scene constraints, and the multi-route nature of trajectory predictions.
Original language | English |
---|---|
Qualification | Doctor of Philosophy |
Awarding Institution |
|
Supervisors/Advisors |
|
Thesis sponsors | |
Award date | 28 Jul 2020 |
DOIs | |
Publication status | Unpublished - 2020 |