Abstract—Problem of transferring and testing self-driving algorithms developed in virtual environment to a physical environment is explored by transferring a Convolutional Neural Network based self-driving car steering algorithm from virtual environment to physical RC card based environment for validation and testing as a step on the way for full scale self-driving car tests. In the process a novel approach for synthetic training data generation from single camera is developed, thus reducing the real world physical requirements for the algorithm and demonstrating the improved self-driving algorithm development pipeline from fully virtual environments to scaled physical models, to full self-driving cars, potentially leveraging the global developer community for development.
Index Terms—Convolutional neural network, training data generation, self-driving cars, virtual and physical models.
The authors are with the Institute of Electronics and Computer Science, Dzerbenes 14, Riga, Latvia (e-mail: firstname.lastname@example.org, e-mail: email@example.com, firstname.lastname@example.org).
Cite: Nauris Dorbe, Ingars Ribners, and Krisjanis Nesenbergs, "Prospects of Improving the self-Driving Car Development Pipeline: Transfer of Algorithms from Virtual to Physical Environment," International Journal of Machine Learning and Computing vol. 8, no. 2, pp. 181-185, 2018.