arXiv subjects

Brendan Tidd

Publications and source records attributed to Brendan Tidd.

At least 19 recordsLinked to original sources

arXiv2026

It's Not Just More Demos: Counterfactual Action Sensitivity Coverage for Data-Efficient Robust Robot Imitation

Visuomotor imitation learning has demonstrated success for manipulation tasks. However, the trained policies remain brittle to visual `nuisances', with even minor task-preserving variations such as lighting, distractions or changes in colour result in heavy degradation of the trained policy's performance. While increasing data diversity can improve robustness, it is unclear which additional demonstrations are informative for a particular trained policy. We propose Counterfactual Nuisance Behaviour Cloning (CFNBC), an offline data-selection framework for targeted robustness repair. Starting from a nominal policy trained on `clean' demonstrations, CFNBC generates paired clean and nuisance observations that preserve the expert action, then measures \emph{action drift}: the change in the policy's predicted action under a nuisance that should not alter the desired behaviour. This provides a policy-specific sensitivity signal for selecting a compact, response-diverse repair set from a larger candidate pool, without requiring rollout success labels or online policy execution. We show in MuJoCo bimanual cube transfer and SimplerEnv cube stacking that action drift correlates with nuisance-induced failure, and that response-guided repair with only $20$--$30$ selected candidates substantially outperforms matched-budget random selection while approaching the performance of much larger random repair budgets. These results support a data-centric view of robustness repair: the most useful data are not necessarily the most numerous, visually diverse, or obviously difficult, but the examples that cover fragile response modes of the current policy.

cs.RO↗

arXiv2026

AnchorVLA: Anchored Diffusion for Efficient End-to-End Mobile Manipulation

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.

cs.RO↗

arXiv2026

SPREAD: Subspace Representation Distillation for Lifelong Imitation Learning

A key challenge in lifelong imitation learning (LIL) is enabling agents to acquire new skills from expert demonstrations while retaining prior knowledge. This requires preserving the low-dimensional manifolds and geometric structures that underlie task representations across sequential learning. Existing distillation methods, which rely on L2-norm feature matching in raw feature space, are sensitive to noise and high-dimensional variability, often failing to preserve intrinsic task manifolds. To address this, we introduce SPREAD, a geometry-preserving framework that employs singular value decomposition (SVD) to align policy representations across tasks within low-rank subspaces. This alignment maintains the underlying geometry of multimodal features, facilitating stable transfer, robustness, and generalization. Additionally, we propose a confidence-guided distillation strategy that applies a Kullback-Leibler divergence loss restricted to the top-M most confident action samples, emphasizing reliable modes and improving optimization stability. Experiments on the LIBERO, lifelong imitation learning benchmark, show that SPREAD substantially improves knowledge transfer, mitigates catastrophic forgetting, and achieves state-of-the-art performance.

cs.LG↗

arXiv2026

RAPT: Model-Predictive Out-of-Distribution Detection and Failure Diagnosis for Sim-to-Real Humanoid Deployment

Deploying learned control policies is risky because policies that appear robust in simulation can confidently enter out-of-distribution (OOD) states after Sim-to-Real transfer, causing silent failures and potential hardware damage. Existing anomaly detectors often fail to meet the requirements of high-rate control, extremely low false-positive rates, and interpretable failure feedback. We present \textbf{RAPT} (Recurrent Anomaly Probabilistic Trajectory Model), a lightweight, self-supervised $50$\,Hz deployment monitor that learns nominal execution from large-scale simulation and produces calibrated, per-dimension predictive-deviation signals online. RAPT enables OOD detection under strict false-positive constraints while localizing \emph{when and where} real execution departs from nominal behavior. For post-hoc diagnosis, RAPT combines temporal saliency, joint-kinematic summaries, and LLM-based semantic reasoning to classify likely failure causes in a zero-shot setting. In simulation across four Isaac Lab tasks, RAPT improves TPR by $37\%$ over the strongest baseline at $0.5\%$ episode-level FPR; on hardware, it achieves $89\%$ TPR across $78$ trials with fewer false positives than high-frequency-compatible baselines, and reaches $75\%$ semantic failure diagnosis accuracy across $21$ categories on a challenging OOD subset. Project website, code, and videos: https://humphreymunn.github.io/RAPT/.

cs.RO↗

arXiv2025

Scalable Multi-Objective Robot Reinforcement Learning through Gradient Conflict Resolution

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.

cs.RO↗

arXiv2025

Improving Generalization Ability of Robotic Imitation Learning by Resolving Causal Confusion in Observations

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.

cs.RO↗

arXiv2024

Whole-Body Dynamic Throwing with Legged Manipulators

Throwing with a legged robot involves precise coordination of object manipulation and locomotion - crucial for advanced real-world interactions. Most research focuses on either manipulation or locomotion, with minimal exploration of tasks requiring both. This work investigates leveraging all available motors (full-body) over arm-only throwing in legged manipulators. We frame the task as a deep reinforcement learning (RL) objective, optimising throwing accuracy towards any user-commanded target destination and the robot's stability. Evaluations on a humanoid and an armed quadruped in simulation show that full-body throwing improves range, accuracy, and stability by exploiting body momentum, counter-balancing, and full-body dynamics. We introduce an optimised adaptive curriculum to balance throwing accuracy and stability, along with a tailored RL environment setup for efficient learning in sparse-reward conditions. Unlike prior work, our approach generalises to targets in 3D space. We transfer our learned controllers from simulation to a real humanoid platform.

cs.RO↗

arXiv2024

M2Distill: Multi-Modal Distillation for Lifelong Imitation Learning

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.

cs.LG↗

arXiv2024

Shape-Space Deformer: Unified Visuo-Tactile Representations for Robotic Manipulation of Deformable Objects

Accurate modelling of object deformations is crucial for a wide range of robotic manipulation tasks, where interacting with soft or deformable objects is essential. Current methods struggle to generalise to unseen forces or adapt to new objects, limiting their utility in real-world applications. We propose Shape-Space Deformer, a unified representation for encoding a diverse range of object deformations using template augmentation to achieve robust, fine-grained reconstructions that are resilient to outliers and unwanted artefacts. Our method improves generalization to unseen forces and can rapidly adapt to novel objects, significantly outperforming existing approaches. We perform extensive experiments to test a range of force generalisation settings and evaluate our method's ability to reconstruct unseen deformations, demonstrating significant improvements in reconstruction accuracy and robustness. Our approach is suitable for real-time performance, making it ready for downstream manipulation applications.

cs.RO↗

arXiv2024

Alternative Interfaces for Human-initiated Natural Language Communication and Robot-initiated Haptic Feedback: Towards Better Situational Awareness in Human-Robot Collaboration

This article presents an implementation of a natural-language speech interface and a haptic feedback interface that enables a human supervisor to provide guidance to, request information, and receive status updates from a Spot robot. We provide insights gained during preliminary user testing of the interface in a realistic robot exploration scenario.

cs.RO↗

arXiv2023

Robotic Vision for Human-Robot Interaction and Collaboration: A Survey and Systematic Review

Robotic vision for human-robot interaction and collaboration is a critical process for robots to collect and interpret detailed information related to human actions, goals, and preferences, enabling robots to provide more useful services to people. This survey and systematic review presents a comprehensive analysis on robotic vision in human-robot interaction and collaboration over the last 10 years. From a detailed search of 3850 articles, systematic extraction and evaluation was used to identify and explore 310 papers in depth. These papers described robots with some level of autonomy using robotic vision for locomotion, manipulation and/or visual communication to collaborate or interact with people. This paper provides an in-depth analysis of current trends, common domains, methods and procedures, technical processes, data sets and models, experimental testing, sample populations, performance metrics and future challenges. This manuscript found that robotic vision was often used in action and gesture recognition, robot movement in human spaces, object handover and collaborative actions, social communication and learning from demonstration. Few high-impact and novel techniques from the computer vision field had been translated into human-robot interaction and collaboration. Overall, notable advancements have been made on how to develop and deploy robots to assist people.

cs.RO↗

arXiv2023

Learning Visuo-Motor Behaviours for Robot Locomotion Over Difficult Terrain

As mobile robots become useful performing everyday tasks in complex real-world environments, they must be able to traverse a range of difficult terrain types such as stairs, stepping stones, gaps, jumps and narrow passages. This work investigated traversing these types of environments with a bipedal robot (simulation experiments), and a tracked robot (real world). Developing a traditional monolithic controller for traversing all terrain types is challenging, and for large physical robots realistic test facilities are required and safety must be ensured. An alternative is a suite of simple behaviour controllers that can be composed to achieve complex tasks. This work efficiently trained complex behaviours to enable mobile robots to traverse difficult terrain. By minimising retraining as new behaviours became available, robots were able to traverse increasingly complex terrain sets, leading toward the development of scalable behaviour libraries.

cs.RO↗

arXiv2023

Heterogeneous robot teams with unified perception and autonomy: How Team CSIRO Data61 tied for the top score at the DARPA Subterranean Challenge

The DARPA Subterranean Challenge was designed for competitors to develop and deploy teams of autonomous robots to explore difficult unknown underground environments. Categorised in to human-made tunnels, underground urban infrastructure and natural caves, each of these subdomains had many challenging elements for robot perception, locomotion, navigation and autonomy. These included degraded wireless communication, poor visibility due to smoke, narrow passages and doorways, clutter, uneven ground, slippery and loose terrain, stairs, ledges, overhangs, dripping water, and dynamic obstacles that move to block paths among others. In the Final Event of this challenge held in September 2021, the course consisted of all three subdomains. The task was for the robot team to perform a scavenger hunt for a number of pre-defined artefacts within a limited time frame. Only one human supervisor was allowed to communicate with the robots once they were in the course. Points were scored when accurate detections and their locations were communicated back to the scoring server. A total of 8 teams competed in the finals held at the Mega Cavern in Louisville, KY, USA. This article describes the systems deployed by Team CSIRO Data61 that tied for the top score and won second place at the event.

cs.RO↗

arXiv2022

Human-Robot Team Performance Compared to Full Robot Autonomy in 16 Real-World Search and Rescue Missions: Adaptation of the DARPA Subterranean Challenge

Human operators in human-robot teams are commonly perceived to be critical for mission success. To explore the direct and perceived impact of operator input on task success and team performance, 16 real-world missions (10 hrs) were conducted based on the DARPA Subterranean Challenge. These missions were to deploy a heterogeneous team of robots for a search task to locate and identify artifacts such as climbing rope, drills and mannequins representing human survivors. Two conditions were evaluated: human operators that could control the robot team with state-of-the-art autonomy (Human-Robot Team) compared to autonomous missions without human operator input (Robot-Autonomy). Human-Robot Teams were often in directed autonomy mode (70% of mission time), found more items, traversed more distance, covered more unique ground, and had a higher time between safety-related events. Human-Robot Teams were faster at finding the first artifact, but slower to respond to information from the robot team. In routine conditions, scores were comparable for artifacts, distance, and coverage. Reasons for intervention included creating waypoints to prioritise high-yield areas, and to navigate through error-prone spaces. After observing robot autonomy, operators reported increases in robot competency and trust, but that robot behaviour was not always transparent and understandable, even after high mission performance.

cs.RO↗

arXiv2022

Residual Skill Policies: Learning an Adaptable Skill-based Action Space for Reinforcement Learning for Robotics

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.

cs.RO↗

arXiv2022

Learning Setup Policies: Reliable Transition Between Locomotion Behaviours

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%).

cs.RO↗

arXiv2021

Passing Through Narrow Gaps with Deep Reinforcement Learning

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.

cs.RO↗

arXiv2021

Learning When to Switch: Composing Controllers to Traverse a Sequence of Terrain Artifacts

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

cs.RO↗