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 for sketching a plan, a parameter calculator for predicting arguments in the plan, and a code generator to convert the plan into executable robot code. 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.
Model-free reinforcement learning is a promising approach for autonomously solving challenging robotics control problems, but faces exploration difficulty without information of the robot's kinematics and dynamics morphology. The under-exploration of multiple modalities with symmetric states leads to behaviors that are often unnatural and sub-optimal. This issue becomes particularly pronounced in the context of robotic systems with morphological symmetries, such as legged robots for which the resulting asymmetric and aperiodic behaviors compromise performance, robustness, and transferability to real hardware. To mitigate this challenge, we can leverage symmetry to guide and improve the exploration in policy learning via equivariance/invariance constraints. In this paper, we investigate the efficacy of two approaches to incorporate symmetry: modifying the network architectures to be strictly equivariant/invariant, and leveraging data augmentation to approximate equivariant/invariant actor-critics. We implement the methods on challenging loco-manipulation and bipedal locomotion tasks and compare with an unconstrained baseline. We find that the strictly equivariant policy consistently outperforms other methods in sample efficiency and task performance in simulation. In addition, symmetry-incorporated approaches exhibit better gait quality, higher robustness and can be deployed zero-shot in real-world experiments.
In the Industrial Internet of Things (IIoTs) and Ocean of Things (OoTs), the advent of massive intelligent services has imposed stringent requirements on both communication and localization, particularly emphasizing precise localization and channel information. This paper focuses on the challenge of jointly optimizing localization and communication in IoT networks. Departing from the conventional independent noise model used in localization and channel estimation problems, we consider a more realistic model incorporating distance-dependent noise variance, as revealed in recent theoretical analyses and experimental results. The distance-dependent noise introduces unknown noise power and a complex noise model, resulting in an exceptionally challenging non-convex and nonlinear optimization problem. In this study, we address a joint localization and channel estimation problem encompassing distance-dependent noise, unknown channel parameters, and uncertainties in sensor node locations. To surmount the intractable nonlinear and non-convex objective function inherent in the problem, we introduce a variational Bayesian learning-based framework. This framework enables the joint optimization of localization and channel parameters by leveraging an effective approximation to the true posterior distribution. Furthermore, the proposed joint learning algorithm provides an iterative closed-form solution and exhibits superior performance in terms of computational complexity compared to existing algorithms. Computer simulation results demonstrate that the proposed algorithm approaches the performance of the Bayesian Cramer-Rao bound (BCRB), achieves localization performance comparable to the ML-GMP algorithm, and outperforms the other two comparison algorithms.
The emerging immersive and autonomous services have posed stringent requirements on both communications and localization. By considering the great potential of reconfigurable intelligent surface (RIS), this paper focuses on the joint channel estimation and localization for RIS-aided wireless systems. As opposed to existing works that treat channel estimation and localization independently, this paper exploits the intrinsic coupling and nonlinear relationships between the channel parameters and user location for enhancement of both localization and channel reconstruction. By noticing the non-convex, nonlinear objective function and the sparser angle pattern, a variational Bayesian learning-based framework is developed to jointly estimate the channel parameters and user location through leveraging an effective approximation of the posterior distribution. The proposed framework is capable of unifying near-field and far-field scenarios owing to exploitation of sparsity of the angular domain. Since the joint channel and location estimation problem has a closed-form solution in each iteration, our proposed iterative algorithm performs better than the conventional particle swarm optimization (PSO) and maximum likelihood (ML) based ones in terms of computational complexity. Simulations demonstrate that the proposed algorithm almost reaches the Bayesian Cramer-Rao bound (BCRB) and achieves a superior estimation accuracy by comparing to the PSO and the ML algorithms.
Can a quadrupedal robot perform bipedal motions like humans? Although developing human-like behaviors is more often studied on costly bipedal robot platforms, we present a solution over a lightweight quadrupedal robot that unlocks the agility of the quadruped in an upright standing pose and is capable of a variety of human-like motions. Our framework is with a bi-level structure. At the low level is a motion-conditioned control policy that allows the quadrupedal robot to track desired base and front limb movements while balancing on two hind feet. The policy is commanded by a high-level motion generator that gives trajectories of parameterized human-like motions to the robot from multiple modalities of human input. We for the first time demonstrate various bipedal motions on a quadrupedal robot, and showcase interesting human-robot interaction modes including mimicking human videos, following natural language instructions, and physical interaction.
Learning Nash equilibrium (NE) in complex zero-sum games with multi-agent reinforcement learning (MARL) can be extremely computationally expensive. Curriculum learning is an effective way to accelerate learning, but an under-explored dimension for generating a curriculum is the difficulty-to-learn of the subgames -- games induced by starting from a specific state. In this work, we present a novel subgame curriculum learning framework for zero-sum games. It adopts an adaptive initial state distribution by resetting agents to some previously visited states where they can quickly learn to improve performance. Building upon this framework, we derive a subgame selection metric that approximates the squared distance to NE values and further adopt a particle-based state sampler for subgame generation. Integrating these techniques leads to our new algorithm, Subgame Automatic Curriculum Learning (SACL), which is a realization of the subgame curriculum learning framework. SACL can be combined with any MARL algorithm such as MAPPO. Experiments in the particle-world environment and Google Research Football environment show SACL produces much stronger policies than baselines. In the challenging hide-and-seek quadrant environment, SACL produces all four emergent stages and uses only half the samples of MAPPO with self-play. The project website is at https://sites.google.com/view/sacl-rl.
Grounded understanding of natural language in physical scenes can greatly benefit robots that follow human instructions. In object manipulation scenarios, existing end-to-end models are proficient at understanding semantic concepts, but typically cannot handle complex instructions involving spatial relations among multiple objects. which require both reasoning object-level spatial relations and learning precise pixel-level manipulation affordances. We take an initial step to this challenge with a decoupled two-stage solution. In the first stage, we propose an object-centric semantic-spatial reasoner to select which objects are relevant for the language instructed task. The segmentation of selected objects are then fused as additional input to the affordance learning stage. Simply incorporating the inductive bias of relevant objects to a vision-language affordance learning agent can effectively boost its performance in a custom testbed designed for object manipulation with spatial-related language instructions.
We consider the problem of cooperative exploration where multiple robots need to cooperatively explore an unknown region as fast as possible. Multi-agent reinforcement learning (MARL) has recently become a trending paradigm for solving this challenge. However, existing MARL-based methods adopt action-making steps as the metric for exploration efficiency by assuming all the agents are acting in a fully synchronous manner: i.e., every single agent produces an action simultaneously and every single action is executed instantaneously at each time step. Despite its mathematical simplicity, such a synchronous MARL formulation can be problematic for real-world robotic applications. It can be typical that different robots may take slightly different wall-clock times to accomplish an atomic action or even periodically get lost due to hardware issues. Simply waiting for every robot being ready for the next action can be particularly time-inefficient. Therefore, we propose an asynchronous MARL solution, Asynchronous Coordination Explorer (ACE), to tackle this real-world challenge. We first extend a classical MARL algorithm, multi-agent PPO (MAPPO), to the asynchronous setting and additionally apply action-delay randomization to enforce the learned policy to generalize better to varying action delays in the real world. Moreover, each navigation agent is represented as a team-size-invariant CNN-based policy, which greatly benefits real-robot deployment by handling possible robot lost and allows bandwidth-efficient intra-agent communication through low-dimensional CNN features. We first validate our approach in a grid-based scenario. Both simulation and real-robot results show that ACE reduces over 10% actual exploration time compared with classical approaches. We also apply our framework to a high-fidelity visual-based environment, Habitat, achieving 28% improvement in exploration efficiency.
Although terahertz (THz) communications can provide mobile broadband services, it usually has a large path loss and is vulnerable to antenna misalignment. This significantly degrades the reception reliability. To address this issue, the hybrid automatic repeat request (HARQ) is proposed to further enhance the reliability of THz communications. This paper provides an in-depth investigation on the outage performance of two different types of HARQ-aided THz communications, including Type-I HARQ and HARQ with chase combining (HARQ-CC). Moreover, the effects of both fading and stochastic antenna misalignment are considered in this paper. The exact outage probabilities of HARQ-aided THz communications are derived in closed-form, with which the asymptotic outage analysis is enabled to explore helpful insights. In particular, it is revealed that full time diversity can be achieved by using HARQ assisted schemes. Besides, the HARQ-CC-aided scheme performs better than the Type-I HARQ-aided one due to its high diversity combining gain. The analytical results are eventually validated via Monte-Carlo simulations.
It has been a recent trend to leverage the power of supervised learning (SL) towards more effective reinforcement learning (RL) methods. We propose a novel phasic approach by alternating online RL and offline SL for tackling sparse-reward goal-conditioned problems. In the online phase, we perform RL training and collect rollout data while in the offline phase, we perform SL on those successful trajectories from the dataset. To further improve sample efficiency, we adopt additional techniques in the online phase including task reduction to generate more feasible trajectories and a value-difference-based intrinsic reward to alleviate the sparse-reward issue. We call this overall algorithm, PhAsic self-Imitative Reduction (PAIR). PAIR substantially outperforms both non-phasic RL and phasic SL baselines on sparse-reward goal-conditioned robotic control problems, including a challenging stacking task. PAIR is the first RL method that learns to stack 6 cubes with only 0/1 success rewards from scratch.