Researchers have proposed a new rehabilitation design that uses brain signals not to control movement directly, but to decide when and how strongly simple haptic cues should be delivered. Combined with wearable devices based on McKibben artificial muscles, this approach may provide intuitive, low-burden guidance that supports motor learning while reducing overreliance on visual and verbal instructions.

Researchers from Donghua University (Shanghai, China) have developed a novel heterogeneous catalyst, diimine-Ni@PCN-701, by immobilizing diimine-nickel complexes into a zirconium-based metal-organic framework (MOF). This catalyst exhibits exceptional activity for propylene dimerization, achieving a turnover frequency (TOF) of up to 56900 h⁻¹ under mild conditions. It selectively converts propylene into C6 dimers, predominantly 4-methyl-2-pentene (4M2P), with 100% selectivity toward dimers. The enhanced performance arises from the electron-withdrawing effect of the Zr₆O₄(OH)₄ clusters and the steric confinement of the MOF pores, offering a general strategy for designing high-performance heterogeneous olefin oligomerization catalysts.

Researchers from Donghua University (Shanghai, China) have made a theoretical development in relational database normalization by investigating the computational complexity of prime attributes and keys specifically for relation schemes in second normal form (2NF). The study demonstrates that even when a relation scheme is known to be in 2NF, the prime attribute problem remains NP-complete, and the number of all keys can grow exponentially with the number of attributes and functional dependencies. The team also proposes a novel algorithm to find all keys in a 2NF relation scheme, deepening the theoretical understanding of 2NF normalization and providing practical references for database schema design.

Traditional server-based datacenters suffer from low resource utilization, insufficient hardware scalability, insufficient resource usage elasticity, and low fault tolerance. With the ever-increasing network speed and processing power in hardware controllers, physically disaggregating the coupled resources in a server has been a promising trend for building next-generation Datacenters, as it effectively addresses the limitations of server-based datacenters.

Non line of sight (NLOS) imaging achieves imaging of occluded objects by analyzing the reflection and scattering of relay surfaces. Deep learning methods are mainly used to restore two-dimensional hidden scenes and face challenges in reconstructing complex real-world scenes. To solve the problems, a research team led by Hao Liu and Mingliang Xu published their new study in Frontiers of Computer Science on 15 May 2026.