FeelAI-Federated Predictive Models on Edge for the Logistic Industry

Purpose and goal:

This project aims to bring the benefits of Federated Learning (FL) to two important Swedish industrial companies: Volvo Group Trucks Technology (VGTT) and Toyota Material Handling Europe (TMHE) for Predictive Maintenance tasks. FL benefits include reducing the volume of data that needs to be collected in one place and transmitted and preserving privacy.

However, there are specific challenges that we seek to address in this project: data heterogeneity among clients, varying levels of available data across different clients, and the need to explain the outcomes of federated models. Expected results and effects:

The emergence of ML in tackling real-world issues is evident, and Sweden has begun to adopt it to advance digital solutions. Typically, the initial approach involves gathering data centrally to build ML models, which is expensive and time-consuming. This project seeks to progress beyond this by leveraging edge computing and employing Federated models.

This approach eliminates the need for centralized data collection, improving data privacy and security, thus promoting more iterative and agile development, which is critical for advancing digitalization in the industry. Approach and implementation:

FeelAI project comprises four work packages (WPs). WP1 covers the entire project duration and includes management and knowledge-sharing activities. WP2 primarily focuses on establishing access to the data on the edges.

Furthermore, only for the purpose of research and experimentation, we aim to gather data from a number of edges. WP3 constitutes the core scientific component, involving the design and implementation of FL algorithms while addressing associated challenges as presented above. WP4 is concerned with practical deployment and presenting the project result.