Vehicle Velocity Prediction Using Artificial Neural Network and Effect of Real World Signals on Prediction Window 2020-01-0729
Prediction of vehicle velocity is important since it can realize improvements in the fuel economy/energy efficiency, drivability, and safety. Velocity prediction has been addressed in many publications. Several references considered deterministic and stochastic approaches such as Markov chain, autoregressive models, and artificial neural networks. There are numerous new sensor and signal technologies like vehicle-to-vehicle and vehicle-to-infrastructure communication that can be used to obtain inclusive datasets. Using these inclusive datasets of sensors in deep neural networks, high accuracy velocity predictions can be achieved. This research builds upon previous findings that Long Short-Term Memory (LSTM) deep neural networks provide low error velocity prediction. We developed an LSTM deep neural network that uses different groups of datasets collected in Fort Collins, Colorado. Synchronous data was gathered using a test vehicle equipped with sensors to measure ego vehicle position and velocity, ADAS-derived near-neighbor relative position and velocity, and infrastructure-derived transit time and signal phase and timing. The effect of different groups of input datasets on forward velocity prediction windows of 10, 15, 20, and 30 seconds was studied. The developed algorithm was tested on an NVIDIA DRIVE PX2. This research shows that the lowest Mean Absolute Error (MAE) of future velocity prediction is with a fully inclusive dataset in 10-second velocity prediction windows. It was observed that GPS data, current vehicle velocity data, and vehicle-to-infrastructure data were the most influential parameters for prediction accuracy. Additionally, we have demonstrated that the LSTM neural network used for velocity prediction can be implemented in real-time using an NVIDIA DRIVE PX2. Integration of velocity prediction into fuel economy strategies and autonomous vehicle technology have the potential to improve fuel economy and safety. Future work involves demonstrating these two use cases in a physical vehicle using NVIDIA DRIVE PX2.
Citation: Gaikwad, T., Rabinowitz, A., Motallebiaraghi, F., Bradley, T. et al., "Vehicle Velocity Prediction Using Artificial Neural Network and Effect of Real World Signals on Prediction Window," SAE Technical Paper 2020-01-0729, 2020, https://doi.org/10.4271/2020-01-0729. Download Citation
Tushar Gaikwad, Aaron Rabinowitz, Farhang Motallebiaraghi, Thomas Bradley, Zachary Asher, Alvis Fong, Rick Meyer
Western Michigan University, Colorado State University