Safe Worker-collaborative Navigation for Robotic Intralogistics in Infrastructure-free Warehouses – NavWare
Safe Worker-collaborative Navigation for Robotic Intralogistics in Infrastructure-free Warehouses
Safe Worker-collaborative Navigation for Robotic Intralogistics in Infrastructure-free Warehouses.
Challenges and objectives
The Industry 4.0 process is profoundly affecting the industrial structure of warehousing and intralogistics. Although existing warehouse robotization solutions explicitly assume the coexistence of autonomous mobile robots (AMRs) and human workers, they require humans to move carefully and that the current observations of the AMRs match their prior knowledge of the working environment. Therefore, these solutions generally have limited deployment space and high operation and maintenance costs. Thus, it is necessary to research and develop next-generation, more reliable, and intelligent robotic navigation methods to enable the large-scale deployment of affordable warehousing and intralogistics automation solutions. The NavWare project proposes to use data-driven deep learning methods to directly intervene in the navigation layers of the AMR for fast, reliable, and local obstacle avoidance and generalizable global path planning, and ultimately generate safe and collaborative robotic navigation for workers. Compared to existing methods, NavWare-based robotic warehouse navigation can be less expensive to deploy and maintain, while system performance can be better.
The four main scientific and technical achievements pursued by the NavWare project are presented as follows: 1) The first is to make RMAs safer and more collaborative to avoid human workers. 2) The second is to also show collaborative characteristics at the level of global path planning, for which the RMA must intelligently plan its path from one working point to another based on time and warehouse configuration. 3) The third is to integrate the above two into a robotic software system called ROS2, and the latter is more industry-friendly, especially compared to ROS. 4) The last is to create an open testbed with a standardizable evaluation process, not only to evaluate the own system but also to facilitate the comparison of different methods within the community.
/
/
/
The process of Industry 4.0 is profoundly affecting the industrial structure of warehousing and intralogistics. However, it is unrealistic for many companies to build fully-automated warehouses by one-time investment due to the limitation of funds and/or restrictions on land registration, or, simply for some companies, Industry 5.0, which emphasises human participation, is more in line with their vision. Although existing solutions explicitly premise the coexistence of autonomous mobile robots (AMRs) and human workers, their main assumptions still include that humans must move carefully, and that the robot's current observations are able to match its priors about the work environment. Consequently, current robotic solutions usually have limited deployment space and high operation-maintenance costs. Therefore, there is a need to research and develop next-generation, more reliable and intelligent robotic navigation methods to enable large-scale deployment of affordable warehousing and intralogistics automation solutions. NavWare proposes to use data-driven deep learning methods to directly intervene in the AMR's navigation layers for fast and reliable local obstacle avoidance as well as generalizable global path planning, and ultimately generate safe worker-collaborative robot navigation. Compared with existing methods, robotic warehouse navigation based on NavWare may be less expensive to deploy and maintain, while the system performance may be better.
Project coordination
Zhi YAN (Connaissance et Intelligence Artificielle Distribuées - UMR 7533)
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
Partnership
CIAD Connaissance et Intelligence Artificielle Distribuées - UMR 7533
Help of the ANR 238,912 euros
Beginning and duration of the scientific project:
February 2024
- 48 Months
