ViCoS Lab

Adaptive deep perception methods for autonomous surface vehicles

The project’s overarching goal is to develop the next-generation maritime environment perception methods, which will harvest the power of end-to-end trainable deep models. The models will address the challenges essential for safe USV operation like general obstacle detection, long-term tracking with re-identification, implicit detection of hazardous areas and sensor fusion for improved detection. Particular focus will be placed on the adaptivity of the models and self-supervised tuning to new environments. New multisensor datasets are planned to be recorded to facilitate this research.

Learning based navigation

We have been developing autonomous mobile robot navigation methods using deep reinforcement learning. Our methods aim to produce navigation policies which are learned completely in simulation and deployed on real robots.

Robust computer vision methods for autonomous water surface vehicles

The project primary goal is to develop functionalities required for robust autonomous navigation of USVs in uncontrolled environments, primarily relying on the captured visual information. The project focuses on obstacle detection using monocular and stereo systems, development of efficient visual tracking algorithms for marine environments and environment representation through sensor fusion.

Tracking by re-detection

We are designing novel trackers capable of target re-detection. These are so-called long-term trackers, applicable to a broad range od surveillance and interactive systems appropriate for autonomous mobile robots, smart cameras, speaker following in videoconferencing systems, and beyond.

Large-Scale Traffic-Sign Detection and Recognition

We developed a Mask-RCNN-based method for large-scale traffic sign detection. The detector is trained on the proposed DFG dataset with over 7000 images of traffic signs. The dataset consists of 200 traffic sign categories of different interclass variabilty. Accurate detection and localization is achieved with the additional offline tracking of the detection forward and backward in time.

Multimodal perception for maritime robotics

We are developing maritime perception models for obstacle detection based on modalities beyond the visible spectrum. The video shows application of thermal camera to river navigation.

Segmentation-based tracking

We are designing novel deep architectures for visual tracking based on segmentation. We proposed a discriminative single-shot segmentation tracker, D3S, which narrows the gap between visual object tracking and video object segmentation. A single-shot network applies two target models with complementary geometric properties, one invariant to a broad range of ransformations, including non-rigid deformations, outperforming the classical bounding box trackers.

Adaptive deep perception methods for autonomous surface vehicles

The project’s overarching goal is to develop the next-generation maritime environment perception methods, which will harvest the power of end-to-end trainable deep models. The models will address the challenges essential for safe USV operation like general obstacle detection, long-term tracking with re-identification, implicit detection of hazardous areas and sensor fusion for improved detection. Particular focus will be placed on the adaptivity of the models and self-supervised tuning to new environments. New multisensor datasets are planned to be recorded to facilitate this research.

Educational robotics

We are developing an open-software open-hardware low-cost light-weight educational platform for robotic manipulation.

Surface defect detection

We are designing novel deep architectures for visual surface inspection. The developed methods allow specialization for large defect detection such as cracks, as well as smooth deformations on reflective surfaces like dents. The methods are learning-based and are thus robust, run realtime and are applicable to a wide range of real problems. Several of the methods are part of most advances surface inspection commercial systems.

Adaptive deep perception methods for autonomous surface vehicles

Learning based navigation

Robust computer vision methods for autonomous water surface vehicles

Tracking by re-detection

Large-Scale Traffic-Sign Detection and Recognition

Multimodal perception for maritime robotics

Segmentation-based tracking

Adaptive deep perception methods for autonomous surface vehicles

Educational robotics

Surface defect detection

About

Location

News