Generating responsive listener head dynamics with nuanced emotions and expressive reactions is crucial for dialogue modeling in various virtual avatar animations. Previous studies mainly focus on the direct short-term production of listener behavior. They overlook the fine-grained control over motion variations and emotional intensity, especially in long-sequence modeling. Moreover, the lack of long-term and large-scale paired speaker-listener corpora incorporating head dynamics and fine-grained multi-modality annotations limits the application of dialogue modeling. Therefore, we first newly collect a large-scale multi-turn dataset of 3D dyadic conversation containing more than 1.4M valid frames for multi-modal responsive interaction, dubbed ListenerX. Additionally, we propose VividListener, a novel framework enabling fine-grained, expressive, and controllable listener dynamics modeling. This framework leverages multi-modal conditions as guiding principles for fostering coherent interactions between speakers and listeners. Specifically, we design the Responsive Interaction Module (RIM) to adaptively represent the multi-modal interactive embeddings. RIM ensures the listener dynamics achieve fine-grained semantic coordination with textual descriptions and adjustments, while preserving expressive reaction with speaker behavior. Meanwhile, we propose the Emotional Intensity Tags (EIT) for emotion intensity editing with multi-modal information integration, applying to both text descriptions and listener motion amplitude. Extensive experiments conducted on our newly collected ListenerX dataset demonstrate that VividListener achieves state-of-the-art performance, realizing expressive and controllable listener dynamics.

Force control strategies are crucial for robot manipulators to effectively adapt to diverse tasks. However, employing force sensors to measure force/torque information significantly increases the cost of robotic systems. Hence, it is crucial to delve into sensorless force/torque estimation techniques. Given the low precision of traditional force estimation algorithms, this article presents a novel algorithm to improve the torque estimation precision of robot manipulators' joints. First, based on the dynamic model, a finite-time sliding-mode torque observer is designed for the robot manipulators via a step-by-step method. A new Lyapunov function is constructed to prove the finite-time convergence of the closed-loop system. Second, in the presence of sampling noise, the robustness of the proposed finite-time torque observer is analyzed. Finally, simulation and experimental results validate the effectiveness of the proposed algorithm.

The stability of second-order non-smooth control system under denial-of-service (DoS) attack is studied in this paper. Non-smooth control method can be used to stabilize the system states to the equilibrium point in a finite time if there is no DoS attack. However, under DoS attack, the controller will maintain the latest received data, which may lead to local divergence of the system states. By designing a Lyapunov function, it is proven that under energy-constrained DoS attack, the states of closed-loop system can still converge to the equilibrium point in a finite time. The simulation results further validate the effectiveness of the theory.

The consensus problem for second-order multiagent systems based on nonsmooth sampled-data control is considered. First, a continuous-time nonsmooth consensus protocol is proposed, which can realize the consensus of systems in a finite time when the external disturbance is absent. Next, based on the sampled data and the zero-order holder, a new discrete-time nonsmooth protocol is proposed. Considering external disturbances, the explicit relationship between the ultimate boundary of errors of any two agents and the sampling period and external disturbance is given with the Lyapunov method and graph theory, which theoretically shows that the nonsmooth control algorithm has a stronger ability to resist external disturbance than the smooth control algorithm. Finally, a simulation example shows the superiority of the nonsmooth consensus algorithm over a smooth consensus algorithm.

Weile Chen, Huiqiang Jiang, Qianhui Wu, Börje Karlsson, Yi Guan. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

With the development of RGB-D sensors, the highquality color point cloud can be obtained conveniently. Besides the geometrical information of point cloud, the color has great potential to assist the point cloud registration. In this paper, we propose a registration method by adaptively combining with color moment information to improve the registration accuracy. Firstly, three kinds of central moments are used to characterize the color distribution of two point clouds. And then, we build the correspondence between point clouds by dynamically combining the geometric feature and color moment feature of each point, hence the corresponding points satisfy both geometric similarity and color similarity. Finally, for the partial registration problem in practice, we apply the trimmed ICP algorithm framework to calculate the rigid transformation. Experimental results demonstrate that our algorithm is more robust and accurate in dealing with point cloud geometry defects, missing data and poor initial position.

In this paper, we propose a novel point set matching algorithm to improve the matching precision in the presence of non-Gaussian noises and outliers. In our method, a non-second order similarity measure known as Kernel Mean p-Power Error (KMPE) loss is employed as the matching cost function. We introduce a local optimal solution for computing the rigid transform by repeating the correspondence estimation and parameter updating processes. This new algorithm assigns a non-linear distance evaluation in kernel space according to the current estimation of the correspondence to yield a more accurate matching result between two point sets in practice. Experimental results demonstrate that our algorithm is more robust and accurate than the traditional ICP and the state-of-the-art algorithms.