Camera-Based Adaptive Trajectory Guidance via Neural Networks
This work addresses indoor robot navigation for dynamic environments, but it is incremental as it builds on existing neural network and image processing methods.
The paper tackles the problem of navigating an indoor robot in dynamic settings by capturing visual trajectories from streaming image data, resulting in neural networks that predict acceleration and steering commands in real time, with experimental results showing performance comparable to human teleoperation and viability in occluded or low-light conditions.
In this paper, we introduce a novel method to capture visual trajectories for navigating an indoor robot in dynamic settings using streaming image data. First, an image processing pipeline is proposed to accurately segment trajectories from noisy backgrounds. Next, the captured trajectories are used to design, train, and compare two neural network architectures for predicting acceleration and steering commands for a line following robot over a continuous space in real time. Lastly, experimental results demonstrate the performance of the neural networks versus human teleoperation of the robot and the viability of the system in environments with occlusions and/or low-light conditions.