The rising demand for robots in warehouses has highlighted the need for efficient multi-robot algorithms. In response, researchers have focused on Multi-Agent Path Finding (MAPF), which enables multiple agents to calculate conflict-free paths to their individual goals. However, the computation time of conflict-based MAPF algorithms significantly increases as the number of conflicts rises, a common challenge in warehouse environments with narrow passages or corridors. To tackle this issue, this study introduces a new type of conflict called “Overlap Conflict.” Overlap Conflicts occur when an agent stops, causing chain conflicts among subsequent agents traveling in the same direction. When an Overlap Conflict arises, the affected agents are dynamically merged into a single group, shifting the conflicts from an individual level to a group level. If the merged agents find themselves with unreachable goals, they are split back into individual agents to continue calculating paths to their respective destinations. This approach effectively reduces computation time in congested environments, particularly in narrow corridors where alternative routes exist.

A robotic focal plane system using robotic fiber positioners enables multi-object spectroscopy for hundreds to thousands of galaxies by utilizing a dense array of positioners that are closely packed at the focal plane of a telescope. While this dense arrangement increases the number of observations, it also introduces the potential for collisions between adjacent positioners. A fiber positioner is designed similarly to a SCARA robot. It is driven by two series of BLDC motors. Each positioner is manufactured with an outer diameter of 16 mm. It operates within an annular workspace with an outer diameter of 33.6 mm and an inner diameter of 12.8 mm. As these positioners are arranged with a spacing of 16.8 mm, target assignment and motion planning are critical to avoid collisions caused by overlapping workspaces. To address this, we proposed an optimized step choice algorithm using a motion planning method based on optimization with the sequential quadratic programming algorithm. Simulation results demonstrated that paths for all positioners within a tile were successfully generated with a success rate of up to 93.75% across 80 tiles.

In the field of robotics and automation, path planning holds significant potential for optimizing field operations. These operations must cover the work area comprehensively and efficiently with minimal movement. To achieve these goals, coverage path planning (CPP) utilizing the Boustrophedon method is essential. However, in an experimental environment, CPP often results in missed work areas due to cumulative sensor errors and structural inconsistencies. This paper aimed to improve CPP by employing the Douglas-Peucker algorithm to simplify the work path and minimizing missed areas. Additionally, Edge Zone Path method was used to generate edge paths, enhancing safety of the trajectory. For experimental purposes, data were acquired from an actual barn. The work area was divided using three segmentation algorithms. Among these, the Voronoi Segmentation, which demonstrated superior performance, was used to extract the data. Experimental results indicated that the proposed optimized CPP improved path safety by maintaining a safe distance from obstacles during frequent turns. Additionally, the Coverage Ratio increased the coverage area of the autonomous robot by an average of 17% compared to the original CPP. These findings suggest that the proposed method can generate more efficient and safe work paths.

Automated valet parking systems have been researched because they provide a good service condition for autonomous vehicles, with their limited space and unmanned environment. Previous parking algorithms focused on planning a path to a parking space based on geometry. However, this approach only works when the parking space is simple. To make automated parking algorithms useful in different environments, it is crucial to drive a path from the entrance to the target space and plan a safe parking path, taking into account the surrounding vehicles in the parking lot. This study organizes the structure of the automated valet parking system into two phases. The first phase involves driving from the origin to the destination. The second phase focuses on planning a path for parking the vehicle in the parking lot. It considers the position, orientation, and parking space to plan a path that aligns correctly. Simulation results demonstrate that the proposed algorithm can plan paths in various parking environments and park vehicles in narrow parking spaces. It is expected that this proposed automated valet parking algorithm can be further improved to contribute to the early commercialization of automated driving technology.