Deep Sensor Fusion for Real-Time Odometry Estimation
This work addresses the calibration-free sensor fusion challenge for low-cost robot navigation, representing an incremental advance by applying CNNs to a known bottleneck in multi-sensor odometry.
The paper tackles the problem of odometry estimation for robot navigation by fusing 2D laser scanners and mono-cameras without needing sensor calibration, using a CNN-based framework that transforms the task into an ordinal classification problem. Results on a real road dataset show real-time performance and improved accuracy over single-sensor methods.
Cameras and 2D laser scanners, in combination, are able to provide low-cost, light-weight and accurate solutions, which make their fusion well-suited for many robot navigation tasks. However, correct data fusion depends on precise calibration of the rigid body transform between the sensors. In this paper we present the first framework that makes use of Convolutional Neural Networks (CNNs) for odometry estimation fusing 2D laser scanners and mono-cameras. The use of CNNs provides the tools to not only extract the features from the two sensors, but also to fuse and match them without needing a calibration between the sensors. We transform the odometry estimation into an ordinal classification problem in order to find accurate rotation and translation values between consecutive frames. Results on a real road dataset show that the fusion network runs in real-time and is able to improve the odometry estimation of a single sensor alone by learning how to fuse two different types of data information.