Find your own way: Weakly-supervised segmentation of path proposals for urban autonomy

By Dan Barnes, William P. Maddern, Ingmar Posner. 2016

134 citations recorded by the source corpus.

We present a weakly-supervised approach to segmenting proposed drivable paths in images with the goal of autonomous driving in complex urban environments. Using recorded routes from a data collection vehicle, our proposed method generates vast quantities of labelled images containing proposed paths and obstacles without requiring manual annotation, which we then use to train a deep semantic segmentation network. With the trained network we can segment proposed paths and obstacles at run-time using a vehicle equipped with only a monocular camera without relying on explicit modelling of road or lane markings. We evaluate our method on the large-scale KITTI and Oxford RobotCar datasets and demonstrate reliable path proposal and obstacle segmentation in a wide variety of environments under a range of lighting, weather and traffic conditions. We illustrate how the method can generalise to multiple path proposals at intersections and outline plans to incorporate the system into a framework for autonomous urban driving.

Subfields: Automotive Engineering, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Automotive Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

Self-Supervised Deep Reinforcement Learning with Generalized Computation Graphs for Robot Navigation (2017, methodology): “These supervised learning methods include learning: drivable routes and then using a planner [13], near-to-far obstacle detectors [14], reactive controllers on top of a map-based planner [15], driving affor-”
A Survey of Autonomous Driving: Common Practices and Emerging Technologies (2019, neutral): “An interesting take on the subject is to segment image data with path proposals using a deep segmentation network [261].”
Composable Action-Conditioned Predictors: Flexible Off-Policy Learning for Robot Navigation (2018, background): “Supervised learning for navigation using off-policy data has been investigated, including learning drivable routes [24] and near-to-far obstacle detectors [25].”
Planning and Decision-Making for Autonomous Vehicles (2018, methodology): “Similarly, a semantic segmentation network may be employed to generate path proposals in the camera image space (69).”
Survey on semantic segmentation using deep learning techniques (2019, methodology): “Segmenting Path Proposals [128] Weakly-supervised approach to segmenting proposed paths for a road vehicle Method for generating a large amount of labeled images without any manual annotation.”
Aggressive Deep Driving: Combining Convolutional Neural Networks and Model Predictive Control (2017, background): “In [19], a neural network is able to produce an image plane driveability map.”
Incremental Adversarial Domain Adaptation for Continually Changing Environments (2017, methodology): “The employed path segmentation labels are generated in a self-supervised setting based on [26].”
Where Should I Walk? Predicting Terrain Properties From Images Via Self-Supervised Learning (2019, unknown): “Weakly and self-supervised learning has increasingly been investigated to obtain scalable navigation solutions, obtaining training data from additional sensors [9], [10] or robot”
Self-Supervised Drivable Area and Road Anomaly Segmentation Using RGB-D Data For Robotic Wheelchairs (2019, methodology): “Barnes et al. [18] proposed an approach to generate training data for the segmentation problem of path proposals by exploiting the information of odometry and obstacle sensing.”
A Survey of State-Action Representations for Autonomous Driving (2018, methodology): “Instead of using the raw images, a preprocessing step can be considered, usually for the sake of semantic segmentation to identify the drivable area of the image, like in (Hadsell et al., 2009; Barnes et al., 2017).”