LiftMeUp · Globally optimal algorithms for dexterous manipulation and locomotion
Horizon Europe — Marie Skłodowska-Curie Actions
- Duration
- 2025-05-01 → 2027-04-30
- EU contribution
- €226,421
- Participants
- 1
- Scheme
- HORIZON-TMA-MSCA-PF-EF
Lines connect the coordinator with its partners.
Results in brief
LiftMeUp: Globally optimal algorithms for dexterous manipulation and locomotion
The goal of LiftMeUp was to develop new algorithms to make robots safer, more efficient, and more trustworthy when performing demanding tasks such as dexterous manipulation and reliable locomotion; capabilities that underpin robots’ potential to operate beyond human limits in domains that require scalable, dependable automation. Today’s mainstream options force a trade-off: first-principles, physics-based methods can be transparent but often require extensive tuning and manual heuristics to perform well, while deep-learning approaches are typically data-hungry, less interpretable, and can generalise poorly. Within the strategic context of the EU’s digital transition and increasing emphasis on AI that is reliable and explainable, the project’s overall objective was to create an easy-to-use framework that uniquely combines data-driven modelling with globally optimal (certifiable) solvers. It aimed to deliver methods that are transparent and sample-efficient, and that reduce sensitivity to initialisation compared with local solvers, which is important for robust performance, safety assurances, and energy/time efficiency. Methodologically, LiftMeUp progressed in three stages: (1) integrating concepts from Koopman theory, polynomial optimisation, and kernel methods to learn “lifting” functions from data and embed them into globally optimal estimation and control; (2) optimally combining different models into more versatile and expressive solutions that can be updated online; and (3) implementing these algorithms on hardware for real-world locomotion and manipulation tasks. The expected impact is both scientific (new links between machine learning and global optimisation) and practical: more dependable robotics that can be deployed with greater confidence, lower data requirements, and improved resource efficiency, supporting broader uptake of advanced robotics in Europe.
Data: CORDIS, © European Union
Project objective
Robots bear the potential to help solve the world’s pressing problems by enabling and scaling up operations beyond human capacities. To successfully manipulate objects and perform reliable locomotion, robots require adequate models and solvers. Traditionally, physics-based models and iterative solvers are used, and obtaining reliable solutions requires significant effort in model tuning and heuristics for good convergence. LiftMeUp’s objective is to combine data-driven modeling with globally optimal solvers in a unique way to create an easy-to-use framework for the life-long operation of robots in challenging tasks. The result is a transparent, sample-efficient alternative to the less interpretable and resource-hungry deep-learning solutions for robotics. Furthermore, LiftMeUp builds on providing certifiably optimal methods with important consequences for safety and efficiency, as opposed to deep learning and local solvers, where different initializations can lead to entirely different solutions. LiftMeUp is carried out at WILLOW, Inria Paris, known for cutting-edge control and locomotion research, and has three stages: first, combining concepts from Koopman theory, polynomial optimization, and kernel methods, lifting functions are inferred from data and integrated into globally optimal methods for state estimation and control. Second, different models are optimally combined, leading to a modular framework that can be incrementally updated online. Lastly, these novel algorithms are implemented on hardware to solve real-world locomotion and dexterous manipulation tasks. This framework will have an important scientific impact by creating novel connections between global optimization and machine learning, enabling the use of principled over heuristic solvers in a broad range of applications in robotics and beyond. It will entail energy and time savings for the economy and using sample-efficient and transparent models will democratize technology and build trust.
Original text from CORDIS.
Participants
- INSTITUT NATIONAL DE RECHERCHE EN INFORMATIQUE ET AUTOMATIQUE · Le Chesnay CedexCoordinatorFrance
Links
- View on CORDIS
- DOI: 10.3030/101207106
- https://ec.europa.eu/research/participants/documents/downloadPublic?documentIds=080166e525395056&appId=PPGMS
Data: CORDIS, © European Union