Tuning swarm behavior for environmental sensing tasks represented as coverage problems

Abpeikar, Shadi; Kasmarik, Kathryn; Tran, Phi Vu; Garratt, Matthew A.; Anavatti, Sreenatha G.; Khan, Md Mohiuddin

doi:10.1016/b978-0-323-90508-4.00001-0

Cited by 7 publications

(6 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It will investigate the changes in the distribution of Sphero BOLT data and how it could be mapped to the observation space of Boids. As discussed in Section 3.1, and in the work done by Abpeikar et al ( 2022c ), the observation space of CoMoT uses configuration parameters of S for a specific number of timesteps and extracts the temporal state values of all the agents during these timesteps. Then the knowledge base of CoMoT on these temporal state values could result in a reward or a penalty for the current state and action.…”

Section: Experiments and Discussionmentioning

confidence: 99%

“…To evaluate the tuned behaviors, the reward value of CoMoT with one shot and iterative transfer learning has been investigated. As mentioned in Abpeikar et al ( 2022a , c ), a cumulative reward signal in the range of [250, 500] indicates a likely collective behavior, while a higher reward indicates more accurate and reliable collective behavior.…”

Section: Experiments and Discussionmentioning

confidence: 99%

“…For the Pioneer 3DX robots, we use the results from Abpeikar et al ( 2022a , c ) for the baseline “without transfer learning” case. Figure 8 , compares the reward signal of CoMoT with the two approaches of transfer learning with that of Abpeikar et al ( 2022a , c ) without any transfer learning. This figure shows the reward signal for collective motion tuning of Pioneer 3DX robots increased when applying the transfer learning.…”

Section: Experiments and Discussionmentioning

confidence: 99%

“…It takes an observed state of the environment as input in each iteration, then selects an action from the action space, which maximizes the corresponding reward (Sutton and Barto, 2018 ). Using human-labeled data as a knowledge base for reward signal generation enables RL to do automatic collective motion tuning (CoMoT) in simulated robots (Abpeikar et al, 2022a ), and collective motion tuning for environmental sensing (Abpeikar et al, 2022c ). However, these methods are only applicable on simulated robots.…”

Section: Background and Related Workmentioning

confidence: 99%

See 3 more Smart Citations

Iterative transfer learning for automatic collective motion tuning on multiple robot platforms

2023

Self Cite

View full text Add to dashboard Cite

This paper proposes an iterative transfer learning approach to achieve swarming collective motion in groups of mobile robots. By applying transfer learning, a deep learner capable of recognizing swarming collective motion can use its knowledge to tune stable collective motion behaviors across multiple robot platforms. The transfer learner requires only a small set of initial training data from each robot platform, and this data can be collected from random movements. The transfer learner then progressively updates its own knowledge base with an iterative approach. This transfer learning eliminates the cost of extensive training data collection and the risk of trial-and-error learning on robot hardware. We test this approach on two robot platforms: simulated Pioneer 3DX robots and real Sphero BOLT robots. The transfer learning approach enables both platforms to automatically tune stable collective behaviors. Using the knowledge-base library the tuning procedure is fast and accurate. We demonstrate that these tuned behaviors can be used for typical multi-robot tasks such as coverage, even though they are not specifically designed for coverage tasks.

show abstract

Section: Experiments and Discussionmentioning

confidence: 99%

Section: Experiments and Discussionmentioning

confidence: 99%

Section: Experiments and Discussionmentioning

confidence: 99%

Section: Background and Related Workmentioning

confidence: 99%

See 2 more Smart Citations

Iterative transfer learning for automatic collective motion tuning on multiple robot platforms

2023

Self Cite

View full text Add to dashboard Cite

show abstract

“…However, based on different scenarios, these distributed tasks may vary. Some examples of distributed tasks include forest fire monitoring and detection [4][5][6], environmental monitoring [7,8], pollution source monitoring and control [9,10], and environmental coverage [11,12]. However, multirobot systems are also robust and scalable for area coverage.…”

Section: Introductionmentioning

confidence: 99%

Distributed Weighted Coverage for Multi-Robot Systems in Non-Convex Environment

et al. 2023

View full text Add to dashboard Cite

Multi-robot coverage systems are widely used in operations such as environmental monitoring, disaster rescue, and pollution prevention. This study considers inherent positioning errors in positioning systems and ground mobile robots with limited communication distance and poor quality in practice. A centroidal Voronoi tessellation algorithm-based formation control technology for multi-robots is optimized. First, by constructing buffered Voronoi cells (BUVCs) for each robot, the collision avoidance ability of the multi-robot formation movement is improved. Next, the formation control problem of multi-robots in a limited communication range and non-convex environment is realized via discrete Voronoi partitioning, a communication distance constraint, and an obstacle avoidance strategy. Simulation and experiment results demonstrate that the proposed method can effectively solve the position generation problem of multi-robot coverage systems in a non-convex environment with actual sizes of the robots and positioning system errors and can further improve the collision avoidance performance of robots and the robustness of BUVC algorithms.

show abstract

Active Sensing Strategy: Multi‐Modal, Multi‐Robot Source Localization and Mapping in Real‐World Settings With Fixed One‐Way Switching

Tran,

Perera,

Garratt

et al. 2024

Journal of Field Robotics

View full text Add to dashboard Cite

This paper introduces a state‐machine model designed for a multi‐modal, multi‐robot environmental sensing algorithm tailored to dynamic real‐world settings. The multi‐modal algorithm uniquely combines two distinct exploration strategies for gas source localization and mapping tasks: (1) an initial exploration phase using multi‐robot coverage path planning with variable formations, providing early gas field indication; and (2) a subsequent active sensing phase employing multi‐robot swarms for precise field estimation. The state machine provides the logic for the transition between these two sensing algorithms. In the exploration phase, a coverage path is generated, maximizing the visited area while measuring gas concentration and estimating the initial gas field at pre‐defined sample times. Subsequently, in the active sensing phase, mobile robots moving in a swarm collaborate to select the next measurement point by broadcasting potential positions and reward values, ensuring coordinated and efficient sensing for a multi‐robot swarm system. System validation involves hardware‐in‐the‐loop experiments and real‐time experiments with a radio source emulating a gas field. The proposed approach is rigorously benchmarked against state‐of‐the‐art single‐mode active sensing and gas source localization techniques. The comprehensive evaluation highlights the multi‐modal switching approach's capacity to expedite convergence, adeptly navigate obstacles in dynamic environments, and significantly enhance the accuracy of gas source location predictions. These findings highlight the effectiveness of our approach, showing significant improvements: a 43% reduction in turnaround time, a 50% increase in estimation accuracy, and enhanced robustness of multi‐robot environmental sensing in cluttered scenarios without collisions. These advancements surpass the performance of conventional active sensing strategies, the partial differential equation model, and geometrical localization approaches, underscoring the efficacy of our method.

show abstract

Tuning swarm behavior for environmental sensing tasks represented as coverage problems

Cited by 7 publications

References 28 publications

Iterative transfer learning for automatic collective motion tuning on multiple robot platforms

Iterative transfer learning for automatic collective motion tuning on multiple robot platforms

Distributed Weighted Coverage for Multi-Robot Systems in Non-Convex Environment

Active Sensing Strategy: Multi‐Modal, Multi‐Robot Source Localization and Mapping in Real‐World Settings With Fixed One‐Way Switching

Contact Info

Product

Resources

About