Pedestrian intention recognition by means of a Hidden Markov Model and body language

According to several reports published by world-wide organisations, thousands of pedestrians die on road accidents every year. Due to this fact, vehicular technologies have been evolving with the intent of reducing these fatalities. Improving these technological advances is crucial since an early recognition of pedestrian intentions can lead to much more accurate active interventions in last second automatic manoeuvres. This paper proposes a method based on a Hidden Markov Model that recognises intentions by means of 3D positions and displacements of 11 joints located along the pedestrian body. The method is able to recognise the intention with an accuracy of 95.13%. It recognises starting intentions 125 ms after gait initiation with an accuracy of 80% and stopping intentions 291.67 and 58.33 ms before the event with an accuracy of 50% and 70% respectively. In addition, an approach based on point clouds and anthropometric constraints to extract the joints in realistic environments is proposed.