We present a large language model (LLM) based system to empower quadrupedal robots with problem-solving abilities for long-horizon tasks beyond short-term motions. Long-horizon tasks for quadrupeds are challenging since they require both a high-level understanding of the semantics of the problem for task planning and a broad range of locomotion and manipulation skills to interact with the environment. Our system builds a high-level reasoning layer with large language models, which generates hybrid discrete-continuous plans as robot code from task descriptions. It comprises multiple LLM agents: a semantic planner that sketches a plan, a parameter calculator that predicts arguments in the plan, a code generator that converts the plan into executable robot code, and a replanner that handles execution failures or human interventions. At the low level, we adopt reinforcement learning to train a set of motion planning and control skills to unleash the flexibility of quadrupeds for rich environment interactions. Our system is tested on long-horizon tasks that are infeasible to complete with one single skill. Simulation and real-world experiments show that it successfully figures out multi-step strategies and demonstrates non-trivial behaviors, including building tools or notifying a human for help. Demos are available on our project page: https://sites.google.com/view/long-horizon-robot.

In-betweening is a technique for generating transitions given start and target character states. The majority of existing works require multiple (often 10) frames as input, which are not always available. In addition, they produce results that lack diversity, which may not fulfill artists' requirements. Addressing these gaps, our work deals with a focused yet challenging problem: generating diverse and high-quality transitions given exactly two frames (only the start and target frames). To cope with this challenging scenario, we propose a bi-directional motion generation and stitching scheme which generates forward and backward transitions from the start and target frames with two adversarial autoregressive networks, respectively, and stitches them midway between the start and target frames. In contrast to stitching at the start or target frames, where the ground truth cannot be altered, there is no strict midway ground truth. Thus, our method can capitalize on this flexibility and generate high-quality and diverse transitions simultaneously. Specifically, we employ conditional variational autoencoders (CVAEs) to implement our autoregressive networks and propose a novel stitching loss to stitch the bi-directional generated motions around the midway point. Extensive experiments demonstrate that our method achieves higher motion quality and more diverse results than existing methods on the LaFAN1, Human3.6m and AMASS datasets.

We aim to control a robot to physically behave in the real world following any high-level language command like "cartwheel" or "kick". Although human motion datasets exist, this task remains particularly challenging since generative models can produce physically unrealistic motions, which will be more severe for robots due to different body structures and physical properties. Deploying such a motion to a physical robot can cause even greater difficulties due to the sim2real gap. We develop LAnguage-Guided mOtion cONtrol (LAGOON), a multi-phase reinforcement learning (RL) method to generate physically realistic robot motions under language commands. LAGOON first leverages a pretrained model to generate a human motion from a language command. Then an RL phase trains a control policy in simulation to mimic the generated human motion. Finally, with domain randomization, our learned policy can be deployed to a quadrupedal robot, leading to a quadrupedal robot that can take diverse behaviors in the real world under natural language commands.