Brendan Tidd

Krishan Rana, Ming Xu, Brendan Tidd et al.

Skill-based reinforcement learning (RL) has emerged as a promising strategy to leverage prior knowledge for accelerated robot learning. Skills are typically extracted from expert demonstrations and are embedded into a latent space from which they can be sampled as actions by a high-level RL agent. However, this skill space is expansive, and not all skills are relevant for a given robot state, making exploration difficult. Furthermore, the downstream RL agent is limited to learning structurally similar tasks to those used to construct the skill space. We firstly propose accelerating exploration in the skill space using state-conditioned generative models to directly bias the high-level agent towards only sampling skills relevant to a given state based on prior experience. Next, we propose a low-level residual policy for fine-grained skill adaptation enabling downstream RL agents to adapt to unseen task variations. Finally, we validate our approach across four challenging manipulation tasks that differ from those used to build the skill space, demonstrating our ability to learn across task variations while significantly accelerating exploration, outperforming prior works. Code and videos are available on our project website: https://krishanrana.github.io/reskill.

Jia Syuen Lim, Zhizhen Zhang, Peter Bohm et al.

A central challenge in mobile manipulation is preserving multiple plausible action models while remaining reactive during execution. A bottle in a cluttered scene can often be approached and grasped in multiple valid ways. Robust behavior depends on preserving this action diversity while remaining reactive as the scene evolves. Diffusion policies are appealing because they model multimodal action distributions rather than collapsing to one solution. But in practice, full iterative denoising is costly at control time. Action chunking helps amortize inference, yet it also creates partially open-loop behavior, allowing small mismatches to accumulate into drift. We present AnchorVLA, a diffusion-based VLA policy for mobile manipulation built on the core insight that when sampling begins near a plausible solution manifold, extensive denoising is unnecessary to recover multimodal, valid actions. AnchorVLA combines a lightweight VLA adaptation backbone with an anchored diffusion action head, which denoises locally around anchor trajectories using a truncated diffusion schedule. This retains multimodal action generation while reducing inference cost for closed-loop control. Crucially, to mitigate chunking-induced drift, we introduce a test-time self-correction mechanism via a lightweight residual correction module that makes high-frequency, per-step adjustments during rollout. Across diverse mobile manipulation tasks, AnchorVLA improves success and stability under disturbances and distribution shifts while maintaining low-latency inference. The source code is made available at https://github.com/jason-lim26/AnchorVLA.

Kaushik Roy, Akila Dissanayake, Brendan Tidd et al.

Lifelong imitation learning for manipulation tasks poses significant challenges due to distribution shifts that occur in incremental learning steps. Existing methods often focus on unsupervised skill discovery to construct an ever-growing skill library or distillation from multiple policies, which can lead to scalability issues as diverse manipulation tasks are continually introduced and may fail to ensure a consistent latent space throughout the learning process, leading to catastrophic forgetting of previously learned skills. In this paper, we introduce M2Distill, a multi-modal distillation-based method for lifelong imitation learning focusing on preserving consistent latent space across vision, language, and action distributions throughout the learning process. By regulating the shifts in latent representations across different modalities from previous to current steps, and reducing discrepancies in Gaussian Mixture Model (GMM) policies between consecutive learning steps, we ensure that the learned policy retains its ability to perform previously learned tasks while seamlessly integrating new skills. Extensive evaluations on the LIBERO lifelong imitation learning benchmark suites, including LIBERO-OBJECT, LIBERO-GOAL, and LIBERO-SPATIAL, demonstrate that our method consistently outperforms prior state-of-the-art methods across all evaluated metrics.

Humphrey Munn, Brendan Tidd, Peter Böhm et al.

Reinforcement Learning (RL) robot controllers usually aggregate many task objectives into one scalar reward. While large-scale proximal policy optimisation (PPO) has enabled impressive results such as robust robot locomotion in the real world, many tasks still require careful reward tuning and are brittle to local optima. Tuning cost and sub-optimality grow with the number of objectives, limiting scalability. Modelling reward vectors and their trade-offs can address these issues; however, multi-objective methods remain underused in RL for robotics because of computational cost and optimisation difficulty. In this work, we investigate the conflict between gradient contributions for each objective that emerge from scalarising the task objectives. In particular, we explicitly address the conflict between task-based rewards and terms that regularise the policy towards realistic behaviour. We propose GCR-PPO, a modification to actor-critic optimisation that decomposes the actor update into objective-wise gradients using a multi-headed critic and resolves conflicts based on the objective priority. Our methodology, GCR-PPO, is evaluated on the well-known IsaacLab manipulation and locomotion benchmarks and additional multi-objective modifications on two related tasks. We show superior scalability compared to parallel PPO (p = 0.04), without significant computational overhead. We also show higher performance with more conflicting tasks. GCR-PPO improves on large-scale PPO with an average improvement of 9.5%, with high-conflict tasks observing a greater improvement. The code is available at https://github.com/humphreymunn/GCR-PPO.

Humphrey Munn, Brendan Tidd, Peter Bohm et al.

Deploying learned control policies on humanoid robots is challenging: policies that appear robust in simulation can execute confidently in out-of-distribution (OOD) states after Sim-to-Real transfer, leading to silent failures that risk hardware damage. Although anomaly detection can mitigate these failures, prior methods are often incompatible with high-rate control, poorly calibrated at the extremely low false-positive rates required for practical deployment, or operate as black boxes that provide a binary stop signal without explaining why the robot drifted from nominal behavior. We present RAPT, a lightweight, self-supervised deployment-time monitor for 50Hz humanoid control. RAPT learns a probabilistic spatio-temporal manifold of nominal execution from simulation and evaluates execution-time predictive deviation as a calibrated, per-dimension signal. This yields (i) reliable online OOD detection under strict false-positive constraints and (ii) a continuous, interpretable measure of Sim-to-Real mismatch that can be tracked over time to quantify how far deployment has drifted from training. Beyond detection, we introduce an automated post-hoc root-cause analysis pipeline that combines gradient-based temporal saliency derived from RAPT's reconstruction objective with LLM-based reasoning conditioned on saliency and joint kinematics to produce semantic failure diagnoses in a zero-shot setting. We evaluate RAPT on a Unitree G1 humanoid across four complex tasks in simulation and on physical hardware. In large-scale simulation, RAPT improves True Positive Rate (TPR) by 37% over the strongest baseline at a fixed episode-level false positive rate of 0.5%. On real-world deployments, RAPT achieves a 12.5% TPR improvement and provides actionable interpretability, reaching 75% root-cause classification accuracy across 16 real-world failures using only proprioceptive data.

Yifei Chen, Yuzhe Zhang, Giovanni D'urso et al.

Recent developments in imitation learning have considerably advanced robotic manipulation. However, current techniques in imitation learning can suffer from poor generalization, limiting performance even under relatively minor domain shifts. In this work, we aim to enhance the generalization capabilities of complex imitation learning algorithms to handle unpredictable changes from the training environments to deployment environments. To avoid confusion caused by observations that are not relevant to the target task, we propose to explicitly learn the causal relationship between observation components and expert actions, employing a framework similar to [6], where a causal structural function is learned by intervention on the imitation learning policy. Disentangling the feature representation from image input as in [6] is hard to satisfy in complex imitation learning process in robotic manipulation, we theoretically clarify that this requirement is not necessary in causal relationship learning. Therefore, we propose a simple causal structure learning framework that can be easily embedded in recent imitation learning architectures, such as the Action Chunking Transformer [31]. We demonstrate our approach using a simulation of the ALOHA [31] bimanual robot arms in Mujoco, and show that the method can considerably mitigate the generalization problem of existing complex imitation learning algorithms.

Nicolas Hudson, Fletcher Talbot, Mark Cox et al.

Heterogeneous teams of robots, leveraging a balance between autonomy and human interaction, bring powerful capabilities to the problem of exploring dangerous, unstructured subterranean environments. Here we describe the solution developed by Team CSIRO Data61, consisting of CSIRO, Emesent and Georgia Tech, during the DARPA Subterranean Challenge. These presented systems were fielded in the Tunnel Circuit in August 2019, the Urban Circuit in February 2020, and in our own Cave event, conducted in September 2020. A unique capability of the fielded team is the homogeneous sensing of the platforms utilised, which is leveraged to obtain a decentralised multi-agent SLAM solution on each platform (both ground agents and UAVs) using peer-to-peer communications. This enabled a shift in focus from constructing a pervasive communications network to relying on multi-agent autonomy, motivated by experiences in early circuit events. These experiences also showed the surprising capability of rugged tracked platforms for challenging terrain, which in turn led to the heterogeneous team structure based on a BIA5 OzBot Titan ground robot and an Emesent Hovermap UAV, supplemented by smaller tracked or legged ground robots. The ground agents use a common CatPack perception module, which allowed reuse of the perception and autonomy stack across all ground agents with minimal adaptation.

Brendan Tidd, Akansel Cosgun, Jurgen Leitner et al.

The U.S. Defense Advanced Research Projects Agency (DARPA) Subterranean Challenge requires teams of robots to traverse difficult and diverse underground environments. Traversing small gaps is one of the challenging scenarios that robots encounter. Imperfect sensor information makes it difficult for classical navigation methods, where behaviours require significant manual fine tuning. In this paper we present a deep reinforcement learning method for autonomously navigating through small gaps, where contact between the robot and the gap may be required. We first learn a gap behaviour policy to get through small gaps (only centimeters wider than the robot). We then learn a goal-conditioned behaviour selection policy that determines when to activate the gap behaviour policy. We train our policies in simulation and demonstrate their effectiveness with a large tracked robot in simulation and on the real platform. In simulation experiments, our approach achieves 93\% success rate when the gap behaviour is activated manually by an operator, and 63\% with autonomous activation using the behaviour selection policy. In real robot experiments, our approach achieves a success rate of 73\% with manual activation, and 40\% with autonomous behaviour selection. While we show the feasibility of our approach in simulation, the difference in performance between simulated and real world scenarios highlight the difficulty of direct sim-to-real transfer for deep reinforcement learning policies. In both the simulated and real world environments alternative methods were unable to traverse the gap.

Brendan Tidd, Nicolas Hudson, Akansel Cosgun et al.

Dynamic platforms that operate over many unique terrain conditions typically require many behaviours. To transition safely, there must be an overlap of states between adjacent controllers. We develop a novel method for training setup policies that bridge the trajectories between pre-trained Deep Reinforcement Learning (DRL) policies. We demonstrate our method with a simulated biped traversing a difficult jump terrain, where a single policy fails to learn the task, and switching between pre-trained policies without setup policies also fails. We perform an ablation of key components of our system, and show that our method outperforms others that learn transition policies. We demonstrate our method with several difficult and diverse terrain types, and show that we can use setup policies as part of a modular control suite to successfully traverse a sequence of complex terrains. We show that using setup policies improves the success rate for traversing a single difficult jump terrain (from 51.3% success rate with the best comparative method to 82.2%), and traversing a random sequence of difficult obstacles (from 1.9% without setup policies to 71.2%).

Brendan Tidd, Nicolas Hudson, Akansel Cosgun et al.

Legged robots often use separate control policiesthat are highly engineered for traversing difficult terrain suchas stairs, gaps, and steps, where switching between policies isonly possible when the robot is in a region that is commonto adjacent controllers. Deep Reinforcement Learning (DRL)is a promising alternative to hand-crafted control design,though typically requires the full set of test conditions to beknown before training. DRL policies can result in complex(often unrealistic) behaviours that have few or no overlappingregions between adjacent policies, making it difficult to switchbehaviours. In this work we develop multiple DRL policieswith Curriculum Learning (CL), each that can traverse asingle respective terrain condition, while ensuring an overlapbetween policies. We then train a network for each destinationpolicy that estimates the likelihood of successfully switchingfrom any other policy. We evaluate our switching methodon a previously unseen combination of terrain artifacts andshow that it performs better than heuristic methods. Whileour method is trained on individual terrain types, it performscomparably to a Deep Q Network trained on the full set ofterrain conditions. This approach allows the development ofseparate policies in constrained conditions with embedded priorknowledge about each behaviour, that is scalable to any numberof behaviours, and prepares DRL methods for applications inthe real world

Brendan Tidd, Nicolas Hudson, Akansel Cosgun

Reliable bipedal walking over complex terrain is a challenging problem, using a curriculum can help learning. Curriculum learning is the idea of starting with an achievable version of a task and increasing the difficulty as a success criteria is met. We propose a 3-stage curriculum to train Deep Reinforcement Learning policies for bipedal walking over various challenging terrains. In the first stage, the agent starts on an easy terrain and the terrain difficulty is gradually increased, while forces derived from a target policy are applied to the robot joints and the base. In the second stage, the guiding forces are gradually reduced to zero. Finally, in the third stage, random perturbations with increasing magnitude are applied to the robot base, so the robustness of the policies are improved. In simulation experiments, we show that our approach is effective in learning walking policies, separate from each other, for five terrain types: flat, hurdles, gaps, stairs, and steps. Moreover, we demonstrate that in the absence of human demonstrations, a simple hand designed walking trajectory is a sufficient prior to learn to traverse complex terrain types. In ablation studies, we show that taking out any one of the three stages of the curriculum degrades the learning performance.