索尼

Artificial intelligence (AI) systems now challenge or surpass human experts in many computer games1,2. Physical and real-time sports such as table tennis, however, remain a major open challenge because of their requirements for fast, precise and adversarial interactions near obstacles and at the edge of human reaction time3. Here we present Ace, to our knowledge the first real-world autonomous system competitive with elite human table tennis players. Ace addresses the challenges of physical real-time interaction through a new, high-speed perception system using event-based vision sensors4, and a new control system based on model-free reinforcement learning, as well as state-of-the-art high-speed robot hardware. Evaluated in matches against elite and professional players under official competition rules, Ace achieved several victories and demonstrated consistent returns of high-speed, high-spin shots. These results highlight the potential of physical AI agents to perform complex, real-time interactive tasks, suggesting broader applications in domains requiring fast, precise human-robot interaction.

Flow map models such as Consistency Models (CM) and Mean Flow (MF) enable few-step generation by learning the long jump of the ODE solution of diffusion models, yet training remains unstable, sensitive to hyperparameters, and costly. Initializing from a pre-trained diffusion model helps, but still requires converting infinitesimal steps into a long-jump map, leaving instability unresolved. We introduce *mid-training*, the first concept and practical method that inserts a lightweight intermediate stage between the (diffusion) pre-training and the final flow map training (i.e., post-training) for vision generation. Concretely, *Consistency Mid-Training* (CMT) is a compact and principled stage that trains a model to map points along a solver trajectory from a pre-trained model, starting from a prior sample, directly to the solver-generated clean sample. It yields a trajectory-consistent and stable initialization. This initializer outperforms random and diffusion-based baselines and enables fast, robust convergence without heuristics. Initializing post-training with CMT weights further simplifies flow map learning. Empirically, CMT achieves state-of-the-art two-step FIDs of 1.97 (CIFAR-10), 1.32 (ImageNet $64\times64$), and 1.84 (ImageNet $512\times512$), using up to $98$\% less training data and GPU time than CMs. On ImageNet $256\times256$, it attains 1-step FID 3.34 with $\sim50$\% less training than MF from scratch (FID 3.43). On MSCOCO T2I, CMT reaches the best FID with $\sim47$\% less training. This establishes CMT as a principled, efficient, and general framework for training flow map models. Code and models are available at https://github.com/sony/cmt.

Recent advances in generative foundational models, often termed "world models," have propelled interest in applying them to critical tasks like robotic planning and autonomous system training. For reliable deployment, these models must exhibit high physical fidelity, accurately simulating real-world dynamics. Existing physics-based video benchmarks, however, suffer from entanglement, where a single test simultaneously evaluates multiple physical laws and concepts, fundamentally limiting their diagnostic capability. We introduce WorldBench, a novel video-based benchmark specifically designed for concept-specific, disentangled evaluation, allowing us to rigorously isolate and assess understanding of a single physical concept or law at a time. To make WorldBench comprehensive, we design benchmarks at two different levels: 1) an evaluation of intuitive physical understanding with higher level concepts such as object permanence or scale/perspective, and 2) an evaluation of low-level physical constants and material properties such as friction coefficients or fluid viscosity, allowing to measure excatly how far from reality generated videos are. When SOTA video-based world models are evaluated on WorldBench, we find specific patterns of failure in particular physics concepts, with all tested models lacking the physical consistency required to generate reliable real-world interactions. Through its concept-specific evaluation, WorldBench offers a more nuanced and scalable framework for rigorously evaluating the physical reasoning capabilities of video generation and world models, paving the way for more robust and generalizable world-model-driven learning.

We present 3DScenePrompt, a framework for camera-controllable video generation that maintains scene consistency when extending arbitrary-length input videos along user-specified trajectories. Unlike existing video generative methods limited to conditioning on a single image or just a few frames, we introduce a dual spatio-temporal conditioning strategy that fundamentally rethinks how video models should reference prior content. Our approach conditions on both temporally adjacent frames for motion continuity and spatially adjacent content for scene consistency. However, when generating beyond temporal boundaries, directly using spatially adjacent frames would incorrectly preserve dynamic elements from the past. We address this through introducing a 3D scene memory that represents exclusively the static geometry extracted from the entire input video. To construct this memory, we leverage dynamic SLAM with our newly introduced dynamic masking strategy that explicitly separates static scene geometry from moving elements. The static scene representation can then be projected to any target viewpoint, providing geometrically-consistent warped views that serve as strong spatial prompts while allowing dynamic regions to evolve naturally from temporal context. This enables our model to maintain long-range spatial coherence and precise camera control without sacrificing computational efficiency or motion realism. Extensive experiments demonstrate that our framework significantly outperforms existing methods in scene consistency, camera controllability, and generation quality.

3D exoscopes have been increasingly adopted in otorhinolaryngology. However, conventional passive-polarized 3D displays (PPD) have limitations in vertical viewing angle, which can impair 3D visualization at nonfrontal angles, particularly for surgical assistants. Active-polarized 3D displays (APD) can overcome these limitations. This study aimed to compare the usability and performance of a conventional PPD and a prototype APD using the ORBEYE 4 K 3D exoscope system under viewing conditions unfavorable to the PPD. Twenty-four otorhinolaryngologists participated in the study. A prototype APD and a commercially available PPD were connected in parallel to the ORBEYE. The participants performed a procedural task simulating stapes surgery with the viewing position set 20° above the display center. The task performance was evaluated based on the number of successfully completed procedural tasks. A target-tracking test was performed before and after the procedural task to evaluate ocular fatigue by calculating the slope of the saccadic main sequence. The usability was assessed using a questionnaire. The APD scored higher than the PPD for all questionnaire items. The number of successful procedural tasks was significantly higher in the APD group. With the APD, there was no change in perceived ocular fatigue before and after the procedural task, whereas with the PPD, there was a tendency toward increased fatigue. The APD demonstrated superior usability and task performance compared to a conventional PPD, particularly under vertically displaced viewing conditions. APD may be particularly beneficial for assistants and surgeons working at various levels of the eye.