Enhancing Exploration Algorithms for Navigation with Visual SLAM

Muravyev, Kirill; Bokovoy, Andrey; Yakovlev, Konstantin

doi:10.1007/978-3-030-86855-0_14

Cited by 4 publications

(6 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Frontier-based methods search for frontiers between the free and the unexplored cells on an occupancy grid map and choose one of these frontiers as a goal. A convenient and effective implementation of a frontier-based exploration algorithm is provided in [35].…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Skill Fusion in Hybrid Robotic Framework for Visual Object Goal Navigation

et al. 2023

Self Cite

View full text Add to dashboard Cite

In recent years, Embodied AI has become one of the main topics in robotics. For the agent to operate in human-centric environments, it needs the ability to explore previously unseen areas and to navigate to objects that humans want the agent to interact with. This task, which can be formulated as ObjectGoal Navigation (ObjectNav), is the main focus of this work. To solve this challenging problem, we suggest a hybrid framework consisting of both not-learnable and learnable modules and a switcher between them—SkillFusion. The former are more accurate, while the latter are more robust to sensors’ noise. To mitigate the sim-to-real gap, which often arises with learnable methods, we suggest training them in such a way that they are less environment-dependent. As a result, our method showed top results in both the Habitat simulator and during the evaluations on a real robot. Video and code for our approach can be found on our website: https://github.com/AIRI-Institute/skill-fusion (accessed on 13 July 2023).

show abstract

Section: Related Workmentioning

confidence: 99%

“…As a goal-setting module, we used our implementation of the frontier-based exploration approach, which was proposed in [35]. The algorithm looks for frontiers between the free and the unknown space on a 2D SLAM-built map.…”

Section: Classical Pipeline 41 Exploration Skillmentioning

confidence: 99%

Skill Fusion in Hybrid Robotic Framework for Visual Object Goal Navigation

et al. 2023

Self Cite

View full text Add to dashboard Cite

show abstract

“…Our toolkit was mentioned in works [18,19] as a tool that is able to run FCNNs in real-time on embedded systems. It was also used in [20] when it was a part of the navigation stack for the vision-based exploration of an unknown environment. In that work, tx2_fcnn_node was run in parallel with SLAM, goal setting, and path planning algorithms.…”

Section: Impactmentioning

confidence: 99%

tx2_fcnn_node: An open-source ROS compatible tool for monocular depth reconstruction

2022

Self Cite

View full text Add to dashboard Cite

“…Accurate visual navigation to target objects in unfamiliar environments is crucial for on-board mobile robot systems. In this case, the selection of embodied agent actions can be performed by various methods: classical modular mapbased motion planning algorithms [1], or end-to-end neural network models based on images from on-board cameras and/or image segmentation masks [2]. Classical approaches use separate modules to build a map, explore the environment, select the final goal, and plan a path to the goal.…”

Section: Introductionmentioning

confidence: 99%

“…Contemporary research [3], [12], as well as solutions of such indoor navigation competitions as the Habitat Challenge 1 , shows that the semantic representation of the map plays a key role in increasing the navigation efficiency. Typical approaches map semantic segmentation results using heuristic projection algorithms and noisy information about the agent's position [13].…”

Section: Introductionmentioning

confidence: 99%

Interactive Semantic Map Representation for Skill-Based Visual Object Navigation

Zemskova,

Staroverov,

Muravyev

et al. 2024

IEEE Access

Self Cite

View full text Add to dashboard Cite

Visual object navigation is one of the key tasks in mobile robotics. One of the most important components of this task is the accurate semantic representation of the scene, which is needed to determine and reach a goal object. This paper introduces a new representation of a scene semantic map formed during the embodied agent interaction with the indoor environment. It is based on a neural network method that adjusts the weights of the segmentation model with backpropagation of the predicted fusion loss values during inference on a regular (backward) or delayed (forward) image sequence. We implement this representation into a full-fledged navigation approach called SkillTron. The method can select robot skills from endto-end policies based on reinforcement learning and classic map-based planning methods. The proposed approach makes it possible to form both intermediate goals for robot exploration and the final goal for object navigation. We conduct intensive experiments with the proposed approach in the Habitat environment, demonstrating its significant superiority over state-of-the-art approaches in terms of navigation quality metrics. The developed code and custom datasets are publicly available at github.com/AIRI-Institute/skillfusion.

show abstract

Enhancing Exploration Algorithms for Navigation with Visual SLAM

Cited by 4 publications

References 23 publications

Skill Fusion in Hybrid Robotic Framework for Visual Object Goal Navigation

Skill Fusion in Hybrid Robotic Framework for Visual Object Goal Navigation

tx2_fcnn_node: An open-source ROS compatible tool for monocular depth reconstruction

Interactive Semantic Map Representation for Skill-Based Visual Object Navigation

Contact Info

Product

Resources

About