Projects
Robotic rat SQuRo
By capturing a functional representation of an actual rat (Rattus norvegicus), we developed a small-sized quadruped robotic rat (SQuRo). Its design consists of four limbs and one flexible spine. Based on extracted key movement joints, SQuRo was delicately designed with a relatively elongated slim body (aspect ratio: 3.42) and lighter weight (220g) compared with quadruped robots of the same scale. Furthermore, a control framework for multi-modal motion planning was proposed, and the corresponding control parameters were tuned through optimization with consideration to stability and actuation limits. Results obtained from series of experimental tests reveal that SQuRo achieves superior motion performance compared with existing state-of-the-art small-sized quadruped robots. Remarkably, SQuRo has an extremely small turning radius (0.48 BL) and strong payload capacity (200 g), and it can recover from falls. The work has been published in IEEE Transactions on Robotics and highlighted by IEEE Spectrum. (https://spectrum.ieee.org/robotic-rat-climbs-crawls-turns)
Biomimetic design based on extraction of KMJs
For a small-scale biomimetic robot, it is challenging to mimic animal-like motion with high speed and high flexibility. Our research focuses on all the aspects of the biomimetic design problem from movement mechanism of animal's model to hardware design. In addition, these methods can be extended to other small-sized biomimetic robots.
We proposed key movement joints (KMJs) to capture a functional representation of the rat with a reduced-order model. By extracting the primary KMJs, we determined the number and distribution of robotic joints for the design of a bioinspired spine mechanism.
We designed a linkage-based slider-coupled symmetric swing mechanism, which features a single-input multiple-output structure, allowing it to couple multiple DOFs. Based on kinematic and dynamic analyses, we optimized key dimensions using an interior-point method to make it compact.
Robot-rat behavioral ineraction
The underlying decision-making mechanisms of rats when they interact with each other still remain mysterious for us humans. To clear up such mysteries, we integrated our robotic rat into the interaction context and developed several methods to generate rat-like motions for the robot. By parameterizing the behavior using a probabilistic model and movement characteristics, some typical behaviors are generated through a second-order Fourier series and a BP neural network. By learning the interaction motion features from rats' demonstrations, two robots perform efficient interaction in simulation, which show that the effective interaction duration between robots has an increment of 34% than that between rats.
Automated manipulation of micro-nano objects
Emerging functional materials such as perovskite, graphene, and molybdenum disulfide have attracted the attention of researchers due to their exceptional multi-physical properties (mechanical, electrical, and optical) compared with their counterparts. To comprehensively characterize their multi-physical properties, we designed and integrated a triple-field-coupled 16 DOF nanomanipulation system into a scanning electron microscope (SEM), which enables the simultaneous imaging, characterization, and direct manipulation of nanoscale materials. A single ZnO nanowire with a 50 nm diameter was picked up and automatically assembled on Ag-based electrodes. Interestingly, a near-infrared light absorption property was exhibited due to the surface plasmon resonance (SPR) occurring at the Ag/ZnO heterojunction.
