Learning-based methods for adaptive informative path planning

Robotics and Autonomous Systems, Journal Year: 2024, Volume and Issue: 179, P. 104727 - 104727

Published: June 4, 2024

adaptive informative path planning (AIPP) is important to many robotics applications, enabling mobile robots efficiently collect useful data about initially unknown environments. In addition, learning-based methods are increasingly used in enhance adaptability, versatility, and robustness across diverse complex tasks. Our survey explores research on applying robotic learning AIPP, bridging the gap between these two fields. We begin by providing a unified mathematical problem definition for general AIPP problems. Next, we establish complementary taxonomies of current work from perspectives (i) algorithms (ii) applications. explore synergies, recent trends, highlight benefits frameworks. Finally, discuss key challenges promising future directions enable more generally applicable robust data-gathering systems through learning. provide comprehensive catalog papers reviewed our survey, including publicly available repositories, facilitate studies field.

Language: Английский

---

Contrastive Learning for Enhancing Robust Scene Transfer in Vision-based Agile Flight

Published: May 13, 2024

Language: Английский

---

Robotic Learning for Informative Path Planning

Published: Jan. 1, 2024

Download This Paper Open PDF in Browser Add to My Library Share: Permalink Using these links will ensure access this page indefinitely Copy URL DOI

Language: Английский

---

Decentralized and Asymmetric Multi-Agent Learning in Construction Sites

IEEE Open Journal of Vehicular Technology, Journal Year: 2024, Volume and Issue: 5, P. 1587 - 1599

Published: Jan. 1, 2024

Multi-agent collaboration involves multiple participants working together in a shared environment to achieve common goal. These agents share information, divide tasks, and synchronize their actions. Key aspects of multi agent include coordination, communication, task allocation, cooperation, adaptation, decentralization. On construction sites, surface grading is the process leveling sand piles increase specific area's height. In this scenario, bulldozer grades while dumper allocates piles. Our work aims utilize multi-agent approach enable these vehicles collaborate effectively. To end, we propose decentralized asymmetric learning for sites (DAMALCS). We formulate DAMALCS reduce expected collisions operating vehicles. Therefore, develop two heuristic experts capable achieving joint goal optimally by applying an innovative prioritization method. approach, bulldozer's movements take precedence over dumper's operations, enabling clear path ensure continuous operation both Since heuristics alone are insufficient real-world scenarios, them train AI agents, which proves be highly effective. simultaneously operate within same environment, aiming avoid optimize performance terms time efficiency volume handling. trained evaluated simulation lab experiments, testing under various conditions, such as visual noise localization errors. The results demonstrate that our significantly reduces collision rates

Language: Английский

---

Learning Where to Look: Self-supervised Viewpoint Selection for Active Localization Using Geometrical Information

Lecture notes in computer science, Journal Year: 2024, Volume and Issue: unknown, P. 188 - 205

Published: Oct. 25, 2024

Language: Английский

---