Learning Accurate, Comfortable and Human-like Driving

Abstract

Autonomous vehicles are more likely to be accepted if they drive accurately, comfortably, but also similar to how human drivers would. This is especially true when autonomous and human-driven vehicles need to share the same road. The main research focus thus far, however, is still on improving driving accuracy only. This paper formalizes the three concerns with the aim of accurate, comfortable and human-like driving. Three contributions are made in this paper. First, numerical map data from HERE Technologies are employed for more accurate driving; a set of map features -which are believed to be relevant to driving -are engineered to navigate better. Second, the learning procedure is improved from a pointwise prediction to a sequence-based prediction and passengers' comfort measures are embedded into the learning algorithm. Finally, we take advantage of the advances in adversary learning to learn human-like driving; specifically, the standard L1 or L2 loss is augmented by an adversary loss which is based on a discriminator trained to distinguish between human driving and machine driving. Our model is trained and evaluated on the Drive360 dataset, which features 60 hours and 3000 km of real-world driving data. Extensive experiments show that our driving model is more accurate, more comfortable and behaves more like a human driver than previous methods. The resources of this work will be released on the project page.

A Survey of End-to-End Driving: Architectures and Training Methods

“…Different versions of ResNet [28] trained on ImageNet dataset are commonly used. The pre-trained layers can be fine-tuned for the current task [83], [84].…”

“…Hecker et al [84] state that while simpler approaches (e.g. camera only) allow to study relevant problems in self-driving, fully autonomous cars require the use of detailed maps.…”

“…For this reason, they augment the Drive360 dataset from [8] with 15 measures (affordances) extracted from detailed maps (see Table I.) TABLE I: List of measures (affordances) extracted from HERE Technologies detailed map and used as inputs in [84].…”

“…Vehicle state measurements such as speed and acceleration are usually middle-fused with visual inputs [6], [66] (concatenation in both cases). Hecker et al [8], [84] middle-fused visual temporal features obtained with LSTM [104] modules from camera feeds by concatenating them with features extracted from maps or from GPS coordinates. As an alternative fusion method, element-wise multiplication was used by Kim et al [87] to middle-fuse textual commands with visual information.…”

Robust Navigation for Racing Drones based on Imitation Learning and Modularization

“…To overcome this problem, Loquercio et al [10] proposed to use datasets for ground vehicles to train autonomous drones instead and Gandhi et al [11] developed a method that automatically generates annotated training data. Besides that, reinforcement learning has also become a popular choice for autonomous navigation [12], [13], [14]. However, a common limitation of these reinforcement learning-based approaches is that it may take a prohibitively long time to converge when the possible state and action space is too large.…”