Vehicle Velocity Prediction Using Artificial Neural Networks and Effect of Real-World Signals on Prediction Window
Date of Award
Master of Science in Engineering
Mechanical and Aerospace Engineering
Dr. Zachary Asher
Dr. Richard Meyer
Dr. Alvis Fong
Autonomous vehicles, artificial neural networks, deep learning, velocity prediction, hybrid electric vehicles
Masters Thesis-Open Access
Prediction of vehicle velocity is essential since it can realize improvements in the fuel economy/energy efficiency, drivability, and safety. Many publications address velocity prediction problems, yet there is a need for the understanding effect of different signals for the prediction. There are numerous new sensor and signal technologies like vehicle-to-vehicle and vehicle-to-infrastructure communication that can be used to obtain comprehensive datasets. Several references considered deterministic and stochastic approaches that use the datasets as input to determine future operation predictions. These approaches include different traffic models and artificial neural networks such as Markov chain, nonlinear autoregressive model, Gaussian function, and recurrent neural network. In this research, we developed different neural networks and machine learning algorithms that use different groups of datasets collected in Fort Collins, Colorado. Synchronous data was gathered using a test vehicle equipped with sensors along the drive route. The custom dataset consists of ego vehicle position, current velocity, ADAS-derived near-neighbor relative position, infrastructure-level transit time, and Signal Phase and Timing (SPaT). The effect of different groups of datasets on future velocity prediction windows of 10, 15, 20, and 30 seconds was studied. The results are assessed based on MAE and time shift. This research shows a 10-second prediction horizon that the lowest Mean Absolute Error (MAE) and the time shift of future velocity prediction. GPS, current vehicle velocity, and Signal Phase and Timing (SPaT) were the most influential parameters for prediction accuracy. Artificial neural networks performed better in terms of getting lower MAE while the classical machine learning model performed better in terms of the getting lower time shift.
Gaikwad, Tushar Dnyaneshwar, "Vehicle Velocity Prediction Using Artificial Neural Networks and Effect of Real-World Signals on Prediction Window" (2020). Masters Theses. 5137.
Electrical and Computer Engineering Commons, Navigation, Guidance, Control, and Dynamics Commons