arXiv subjects

Haizhou Zhao

Publications and source records attributed to Haizhou Zhao.

16 recordsLinked to original sources

arXiv2026

MotionDisco: Motion Discovery for Extreme Humanoid Loco-Manipulation

We present MotionDisco, a framework that discovers contact-rich, long-horizon humanoid loco-manipulation motions from scratch, without relying on teleoperation or motion retargeting from human demonstrations. This is challenging because the space of possible contact interactions grows combinatorially with the task horizon and the number of objects in the scene. MotionDisco enables rapid discovery of novel motions by coupling a large language model (LLM) guided evolutionary search over sequences of interactions with an efficient sequential kinodynamic trajectory optimizer and pruning strategy, enabling the rapid discovery of novel skills. Through extensive ablation studies, we show that our LLM-guided search discovers successful whole-body trajectories across several challenging long-horizon tasks. Finally, by training reinforcement learning tracking policies on the discovered trajectories, we transfer the motions to a real humanoid robot. This is the first work to discover and deploy long-horizon humanoid loco-manipulation skills entirely through automated evolutionary search. Supplementary videos of the experiments are available at: https://youtu.be/DHiVz34QYlw.

cs.RO↗

arXiv2026

Hippo: High-performance Interior-Point and Projection-based Solver for Generic Constrained Trajectory Optimization

Trajectory optimization is the core of modern model-based robotic control and motion planning. Existing trajectory optimizers, based on sequential quadratic programming (SQP) or differential dynamic programming (DDP), are often limited by their slow computation efficiency, low modeling flexibility, and poor convergence for complex tasks requiring hard constraints. In this paper, we introduce Hippo, a solver that can handle inequality constraints using the interior-point method (IPM) with an adaptive barrier update strategy and hard equality constraints via projection or IPM. Through extensive numerical benchmarks, we show that Hippo is a robust and efficient alternative to existing state-of-the-art solvers for difficult robotic trajectory optimization problems requiring high-quality solutions, such as locomotion and manipulation.

cs.RO↗

arXiv2026

SURE: Safe Uncertainty-Aware Robot-Environment Interaction using Trajectory Optimization

Robotic tasks involving contact interactions pose significant challenges for trajectory optimization due to discontinuous dynamics. Conventional formulations typically assume deterministic contact events, which limit robustness and adaptability in real-world settings. In this work, we propose SURE, a robust trajectory optimization framework that explicitly accounts for contact timing uncertainty. By allowing multiple trajectories to branch from possible pre-impact states and later rejoin a shared trajectory, SURE achieves both robustness and computational efficiency within a unified optimization framework. We evaluate SURE on two representative tasks with unknown impact times. In a cart-pole balancing task involving uncertain wall location, SURE achieves an average improvement of 21.6% in success rate when branch switching is enabled during control. In an egg-catching experiment using a robotic manipulator, SURE improves the success rate by 40%. These results demonstrate that SURE substantially enhances robustness compared to conventional nominal formulations.

cs.RO↗

arXiv2025

Let It Flow: Agentic Crafting on Rock and Roll, Building the ROME Model within an Open Agentic Learning Ecosystem

Agentic crafting requires LLMs to operate in real-world environments over multiple turns by taking actions, observing outcomes, and iteratively refining artifacts. Despite its importance, the open-source community lacks a principled, end-to-end ecosystem to streamline agent development. We introduce the Agentic Learning Ecosystem (ALE), a foundational infrastructure that optimizes the production pipeline for agentic model. ALE consists of three components: ROLL, a post-training framework for weight optimization; ROCK, a sandbox environment manager for trajectory generation; and iFlow CLI, an agent framework for efficient context engineering. We release ROME, an open-source agent grounded by ALE and trained on over one million trajectories. Our approach includes data composition protocols for synthesizing complex behaviors and a novel policy optimization algorithm, Interaction-Perceptive Agentic Policy Optimization (IPA), which assigns credit over semantic interaction chunks rather than individual tokens to improve long-horizon training stability. Empirically, we evaluate ROME within a structured setting and introduce Terminal Bench Pro, a benchmark with improved scale and contamination control. ROME demonstrates strong performance across benchmarks like SWE-bench Verified and Terminal Bench, proving the effectiveness of ALE.

cs.AI↗

arXiv2025

RollArt: Disaggregated Multi-Task Agentic RL Training at Scale

Agentic Reinforcement Learning (RL) trains LLMs through multi-turn interactions with environments, producing workloads that mix compute-bound prefill, bandwidth-bound decoding, CPU-heavy environment execution, and bursty reward evaluation. Existing systems either colocate all stages on a single GPU cluster or decouple them only at a coarse granularity, overlooking hardware heterogeneity and incurring substantial synchronization overhead across stages. We present ROLLART, a system for multi-task agentic RL on disaggregated infrastructure. ROLLART maps each pipeline stage to best-fit hardware, routing prefill-heavy tasks to compute-optimized GPUs, decode-heavy tasks to bandwidth-optimized GPUs, and environments to CPU clusters. It decouples rollout at the trajectory level, allowing generation, environment interaction, and reward scoring to proceed independently, so that slow or failed environments never block the others. ROLLART offloads stateless reward computation to serverless infrastructure and overlaps rollout with training via staleness-bounded asynchronous weight synchronization. Our results demonstrate that ROLLART effectively improves training throughput and achieves 1.31--2.05 \(\times\) training time reduction compared to various RL systems. We also evaluated ROLLART by training a hundreds-of-billions-parameter MoE model for Qoder product on an Alibaba cluster with above 3,000 GPUs, demonstrating its stability and scalability.

cs.DC↗

arXiv2025

Part II: ROLL Flash -- Accelerating RLVR and Agentic Training with Asynchrony

Synchronous Reinforcement Learning (RL) post-training has emerged as a crucial step for enhancing Large Language Models (LLMs) with diverse capabilities. However, many systems designed to accelerate RL post-training still suffer from low resource utilization and limited scalability. We present ROLL Flash, a system that extends ROLL with native support for asynchronous RL post-training. ROLL Flash is built upon two core design principles: fine-grained parallelism and rollout-train decoupling. Guided by these principles, ROLL Flash provides flexible programming interfaces that enable a fully asynchronous training architecture and support efficient rollout mechanisms, including queue scheduling and environment-level asynchronous execution. Through comprehensive theoretical analysis and extensive experiments, we demonstrate that ROLL Flash significantly improves resource utilization and scalability over synchronous RL post-training. ROLL Flash achieves up to 2.24x speedup on RLVR tasks and 2.72x on agentic tasks, using the same GPU budget as synchronous baselines. Furthermore, we implement several popular off-policy algorithms and verify that asynchronous training can achieve performance on par with synchronous training.

cs.LG↗

arXiv2025

Simultaneous Contact Sequence and Patch Planning for Dynamic Locomotion

Legged robots have the potential to traverse highly constrained environments with agile maneuvers. However, planning such motions requires solving a highly challenging optimization problem with a mixture of continuous and discrete decision variables. In this paper, we present a full pipeline based on Monte-Carlo tree search (MCTS) and whole-body trajectory optimization (TO) to perform simultaneous contact sequence and patch selection on highly challenging environments. Through extensive simulation experiments, we show that our framework can quickly find a diverse set of dynamically consistent plans. We experimentally show that these plans are transferable to a real quadruped robot. We further show that the same framework can find highly complex acyclic humanoid maneuvers. To the best of our knowledge, this is the first demonstration of simultaneous contact sequence and patch selection for acyclic multi-contact locomotion using the whole-body dynamics of a quadruped.

cs.RO↗

arXiv2025

Reinforcement Learning Optimization for Large-Scale Learning: An Efficient and User-Friendly Scaling Library

We introduce ROLL, an efficient, scalable, and user-friendly library designed for Reinforcement Learning Optimization for Large-scale Learning. ROLL caters to three primary user groups: tech pioneers aiming for cost-effective, fault-tolerant large-scale training, developers requiring flexible control over training workflows, and researchers seeking agile experimentation. ROLL is built upon several key modules to serve these user groups effectively. First, a single-controller architecture combined with an abstraction of the parallel worker simplifies the development of the training pipeline. Second, the parallel strategy and data transfer modules enable efficient and scalable training. Third, the rollout scheduler offers fine-grained management of each sample's lifecycle during the rollout stage. Fourth, the environment worker and reward worker support rapid and flexible experimentation with agentic RL algorithms and reward designs. Finally, AutoDeviceMapping allows users to assign resources to different models flexibly across various stages.

cs.LG↗

arXiv2025

Physically Consistent Humanoid Loco-Manipulation using Latent Diffusion Models

This paper uses the capabilities of latent diffusion models (LDMs) to generate realistic RGB human-object interaction scenes to guide humanoid loco-manipulation planning. To do so, we extract from the generated images both the contact locations and robot configurations that are then used inside a whole-body trajectory optimization (TO) formulation to generate physically consistent trajectories for humanoids. We validate our full pipeline in simulation for different long-horizon loco-manipulation scenarios and perform an extensive analysis of the proposed contact and robot configuration extraction pipeline. Our results show that using the information extracted from LDMs, we can generate physically consistent trajectories that require long-horizon reasoning.

cs.RO↗

arXiv2025

Baichuan-M1: Pushing the Medical Capability of Large Language Models

The current generation of large language models (LLMs) is typically designed for broad, general-purpose applications, while domain-specific LLMs, especially in vertical fields like medicine, remain relatively scarce. In particular, the development of highly efficient and practical LLMs for the medical domain is challenging due to the complexity of medical knowledge and the limited availability of high-quality data. To bridge this gap, we introduce Baichuan-M1, a series of large language models specifically optimized for medical applications. Unlike traditional approaches that simply continue pretraining on existing models or apply post-training to a general base model, Baichuan-M1 is trained from scratch with a dedicated focus on enhancing medical capabilities. Our model is trained on 20 trillion tokens and incorporates a range of effective training methods that strike a balance between general capabilities and medical expertise. As a result, Baichuan-M1 not only performs strongly across general domains such as mathematics and coding but also excels in specialized medical fields. We have open-sourced Baichuan-M1-14B, a mini version of our model, which can be accessed through the following links.

cs.CL↗

arXiv2024

Trajectory Optimization under Contact Timing Uncertainties

Most interesting problems in robotics (e.g., locomotion and manipulation) are realized through intermittent contact with the environment. Due to the perception and modeling errors, assuming an exact time for establishing contact with the environment is unrealistic. On the other hand, handling uncertainties in contact timing is notoriously difficult as it gives rise to either handling uncertain complementarity systems or solving combinatorial optimization problems at run-time. This work presents a novel optimal control formulation to find robust control policies under contact timing uncertainties. Our main novelty lies in casting the stochastic problem to a deterministic optimization over the uncertainty set that ensures robustness criterion satisfaction of candidate pre-contact states and optimizes for contact-relevant objectives. This way, we only need to solve a manageable standard nonlinear programming problem without complementarity constraints or combinatorial explosion. Our simulation results on multiple simplified locomotion and manipulation tasks demonstrate the robustness of our uncertainty-aware formulation compared to the nominal optimal control formulation.

cs.RO↗

arXiv2024

Arm-Constrained Curriculum Learning for Loco-Manipulation of the Wheel-Legged Robot

Incorporating a robotic manipulator into a wheel-legged robot enhances its agility and expands its potential for practical applications. However, the presence of potential instability and uncertainties presents additional challenges for control objectives. In this paper, we introduce an arm-constrained curriculum learning architecture to tackle the issues introduced by adding the manipulator. Firstly, we develop an arm-constrained reinforcement learning algorithm to ensure safety and stability in control performance. Additionally, to address discrepancies in reward settings between the arm and the base, we propose a reward-aware curriculum learning method. The policy is first trained in Isaac gym and transferred to the physical robot to do dynamic grasping tasks, including the door-opening task, fan-twitching task and the relay-baton-picking and following task. The results demonstrate that our proposed approach effectively controls the arm-equipped wheel-legged robot to master dynamic grasping skills, allowing it to chase and catch a moving object while in motion. Please refer to our website (https://acodedog.github.io/wheel-legged-loco-manipulation) for the code and supplemental videos.

cs.RO↗

arXiv2023

Baichuan 2: Open Large-scale Language Models

Large language models (LLMs) have demonstrated remarkable performance on a variety of natural language tasks based on just a few examples of natural language instructions, reducing the need for extensive feature engineering. However, most powerful LLMs are closed-source or limited in their capability for languages other than English. In this technical report, we present Baichuan 2, a series of large-scale multilingual language models containing 7 billion and 13 billion parameters, trained from scratch, on 2.6 trillion tokens. Baichuan 2 matches or outperforms other open-source models of similar size on public benchmarks like MMLU, CMMLU, GSM8K, and HumanEval. Furthermore, Baichuan 2 excels in vertical domains such as medicine and law. We will release all pre-training model checkpoints to benefit the research community in better understanding the training dynamics of Baichuan 2.

cs.CL↗

arXiv2023

A Compact Variable Stiffness Actuator for Agile Legged Locomotion

The legged robots with variable stiffness actuators (VSAs) can achieve energy-efficient and versatile locomotion. However, equipping legged robots with VSAs in real-world application is usually restricted by (i) the redundant mechanical structure design, (ii) limited stiffness variation range and speed, (iii) high energy consumption in stiffness modulation, and (iv) the lack of online stiffness control structure with high performance. In this paper, we present a novel Variable-Length Leaf-Spring Actuator (VLLSA) designed for legged robots that aims to address the aforementioned limitations. The design is based on leaf-spring mechanism and we improve the structural design to make the proposed VSA (i) compact and lightweight in mechanical structure, (ii) precise in theoretical modeling, and (iii) capable of modulating stiffness with wide range, fast speed, low energy consumption and high control performance. Hardware experiments including in-place and forward hopping validate advantages of the proposed VLLSA.

cs.RO↗

arXiv2023

Design and Control of a Bio-inspired Wheeled Bipedal Robot

Wheeled bipedal robots (WBRs) have the capability to execute agile and versatile locomotion tasks. This paper focuses on improving the dynamic performance of WBRs through innovations in both hardware and software development. Inspired by the human barbell squat, a bionic mechanical design is proposed and implemented as shown in Fig. 1. It distributes the weight onto its hip and knee joints to improve the effectiveness of joint motors while maintaining a relatively large workspace of the base link. Meanwhile, a novel model-based controller is devised, synthesizing height-variable wheeled linear inverted pendulum (HV-wLIP) model, Control Lyapunov Function (CLF) and whole-body dynamics for theoretically guaranteed stability and efficient computation. Compared with other alternatives, as a more accurate approximation of the WBR dynamics, the HV-wLIP can enable more agile response and provide theory basis for WBR controller design. Experimental results demonstrate that the robot could perform human-like deep squat, and is capable of maintaining tracking CoM velocity while manipulating base states. Furthermore, it exhibited robustness against external disturbances and unknown terrains even in the wild.

cs.RO↗

arXiv2017

Assigning personality/identity to a chatting machine for coherent conversation generation

Endowing a chatbot with personality or an identity is quite challenging but critical to deliver more realistic and natural conversations. In this paper, we address the issue of generating responses that are coherent to a pre-specified agent profile. We design a model consisting of three modules: a profile detector to decide whether a post should be responded using the profile and which key should be addressed, a bidirectional decoder to generate responses forward and backward starting from a selected profile value, and a position detector that predicts a word position from which decoding should start given a selected profile value. We show that general conversation data from social media can be used to generate profile-coherent responses. Manual and automatic evaluation shows that our model can deliver more coherent, natural, and diversified responses.

cs.CL↗