Cyclist Intention Detection: A Probabilistic Approach

TL;DR

This paper introduces a probabilistic ensemble approach for cyclist intention detection using motion history images and residual CNNs to produce reliable, sharp, and accurate trajectory forecasts with Gaussian mixtures.

Contribution

The paper presents a novel probabilistic ensemble method combining specialized models and cyclist motion probabilities for improved trajectory forecasting.

Findings

The ensemble approach yields more reliable forecasts than single models.

Forecasts are sharper and maintain comparable positional accuracy.

Method is validated on real-world traffic intersection data.

Abstract

This article presents a holistic approach for probabilistic cyclist intention detection. A basic movement detection based on motion history images (MHI) and a residual convolutional neural network (ResNet) are used to estimate probabilities for the current cyclist motion state. These probabilities are used as weights in a probabilistic ensemble trajectory forecast. The ensemble consists of specialized models, which produce individual forecasts in the form of Gaussian distributions under the assumption of a certain motion state of the cyclist (e.g. cyclist is starting or turning left). By weighting the specialized models, we create forecasts in the from of Gaussian mixtures that define regions within which the cyclists will reside with a certain probability. To evaluate our method, we rate the reliability, sharpness, and positional accuracy of our forecasted distributions. We compare our method to a single model approach which produces forecasts in the form of Gaussian distributions and show that our method is able to produce more reliable and sharper outputs while retaining comparable positional accuracy. Both methods are evaluated using a dataset created at a public traffic intersection. Our code and the dataset are made publicly available.