On the Road with 16 Neurons: Mental Imagery with Bio-inspired Deep Neural Networks
Submitted to Image and Vision Computing
This paper proposes a strategy for visual perception in the context of autonomous driving. Humans, when not distracted or drunk, are still the best drivers you can currently ﬁnd. For this reason we take inspiration from two theoretical ideas about the human mind and its neural organization. The ﬁrst idea concerns how the brain uses a hierarchical structure of neuron ensembles to extract abstract concepts from visual experience and code them into compact representations. The second idea suggests that these neural perceptual representations are not neutral but functional to the prediction of the future state of aﬀairs in the environment. Similarly, the prediction mechanism is not neutral but oriented to the current planning of a future action. We identify within the deep learning framework two artiﬁcial counterparts of the aforementioned neurocognitive theories. We ﬁnd a correspondence between the ﬁrst theoretical idea and the architecture of convolutional autoencoders, while we translate the second theory into a training procedure that learns compact representations which are not neutral but oriented to driving tasks, from two distinct perspectives. From a static perspective, we force groups of neural units in the compact representations to distinctly represent speciﬁc concepts crucial to the driving task. From a dynamic perspective, we encourage the compact representations to be predictive of how the current road scenario will change in the future. We successfully learn compact representations that use as few as 16 neural units for each of the two basic driving concepts we consider: cars and lanes. We prove the eﬃciency of our proposed perceptual representations on the SYNTHIA dataset.