eess.SYSep 19, 2026

Task aware Dynamic Movement Primitives for failure detection and recovery in contact rich manipulation

Authors: Bhavnashri A, Sobia Shafi, Krishnapuram Himavarshini, Anuj Tiwari

Organizations: Atomberg Technologies Limited, India · Indian Institute of Technology Madras (IITM), India · Appian Corporation, India

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Assembly remains a challenging robotic manipulation task in presence of tight tolerances and complex contact interactions. While Learning from Demonstration (LfD) frameworks like Dynamic Movement Primitives (DMPs) can effectively encode trajectories from a single demonstration, they are highly sensitive to variations in initial grasp configurations and external contact forces. Such variations often lead to task failures during the contact rich phases. This paper presents a task aware failure detection and recovery framework that integrates DMP based trajectory generation with real time stage classification. Utilizing Quadratic Discriminant Analysis (QDA) trained on multimodal sensor data, the framework segments execution into approach, alignment, and insertion stages for a Peg in Hole (PiH) assembly operation. By using goal relative position data as features, this classification generalizes to unseen goal positions without requiring retraining, matching the inherent generalization capability of DMPs. Anomaly detection is performed online using a Mahalanobis distance metric computed over force features, isolating contact induced failures from nominal trajectory execution. Upon failure detection, a spiral search recovery policy is triggered to actively realign the peg under contact before resuming the learned DMP insertion. The proposed approach is evaluated on an experimental setup achieving 95% stage classification accuracy, and demonstrates reliable failure recovery under lateral misalignments of up to 3 mm using only a single demonstration.

Figures & tables

Fig. 1: Overview of proposed learning and recovery framework, starting with 1. single task demonstration through teleoperation, 2. QDA training for stage classification, 3. encoding the trajectory using dynamic movement primitive, and 4. the proposed framework for the Peg in Hole (PiH) task.
Fig. 2: Experimental setup for collecting task demonstrations via teleoperation
Fig. 3: 3D-printed peg in hole insertion setup for demonstrating contact rich task
Fig. 4: Stage prediction using QDA on DMP execution
Fig. 5: Demonstration data with time labeled task stages used for training
TrialAccuracy (%)
197.5
288.7
395.0
486.1
585.4
698.0
TABLE I: Stage Classification Accuracy Across Recorded Execution Trials
StageTrue Time Interval (s)Predicted Time Interval (s)
Approach0.00.0 – 17.0017.000.00.0 – 16.3616.36
Alignment17.0017.00 – 32.8032.8016.3616.36 – 32.3432.34
Insertion32.8032.80 – 38.0038.0032.3432.34 – 38.0038.00
Overall Accuracy95%
TABLE II: Comparison of True and Predicted Stage Time Intervals
Fig. 6: Comparison of demonstration trajectory and DMP execution under a shifted goal position
Fig. 7: Stage classification results on DMP execution with shifted goal position where the classifier correctly generalises across goal variations.
Fig. 8: Comparison of force profiles between demonstration and execution.
Fig. 9: Sharp rise in force along z direction during insertion stage due to failure.
Fig. 10: Mahalanobis distance during execution with six-sigma threshold.
More in this topic

Robotic Manipulation

Robotic manipulation studies how robots perceive, plan, and control physical interactions to grasp, move, assemble, and use objects.

See all 1,685 papers

Explore similar work

CardsList
Oct 1, 2026

Interaction-Stiffness-Guided Basis Allocation in Dynamic Movement Primitives for Efficient Skill Transfer

Dynamic Movement Primitives (DMPs) provide a compact and stable formulation for trajectory representation and generalization in robot skill learning. However, their predefined basis layout limits the allocation of approximation capacity according to stage-dependent precision requirements. To address this issue, this article proposes Stage-Criticality-Guided Dynamic Movement Primitives (SC-DMPs) with adaptive basis allocation for precision-critical skill learning. Operator-robot interaction stiffness and a trajectory-consistency cue derived from cross-demonstration task-space variability are integrated to construct a stage-criticality index. Guided by this index, basis centers are redistributed in normalized time through inverse cumulative criticality and mapped to the canonical phase domain, while their bandwidths are refined to adjust local approximation support. This enables denser and more flexible representation at high-criticality stages while retaining sparser allocation elsewhere. Experiments on handwriting trajectories and three real-robot tasks show that the inferred criticality is concentrated in geometrically demanding and task-constrained regions. Comparisons with DMPs, ProMPs, ProDMP, GP-MP, and KMP demonstrate improved trajectory reproduction, endpoint generalization, and task-critical accuracy while retaining a compact model and the stable structure of classical DMPs.

Chan Xu, Silu Chen, Dehao Wang +6

Zhejiang Key Laboratory of Precision Actuation and Intelligent Robotics, Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences, Ningbo 315201, China · University of Chinese Academy of Sciences, Beijing 100049, China · Department of Computing, The Hong Kong Polytechnic University, HKSAR, China

Robot Motion GenerationRobot Imitation LearningTrajectory Representation LearningRobot Learning
AskPDF
May 17, 2026

From a Single Demonstration to a General Policy for Contact-Rich Manipulation

We present a Learning from Demonstration (LfD) framework that achieves one-shot generalization in multi-stage, contact-rich manipulation tasks. Central to our approach is the utilization of environmental constraints as the inductive bias. By representing a demonstration as a sequence of behaviors that exploit environmental constraints, the robot separates task-general structure -- the constraint types and their transitions -- from instance-specific details such as exact demonstration trajectories, poses, and local geometries. Our four-stage pipeline builds a complete policy on this representation: the robot first abstracts a single demonstration into environmental-constraint primitives, then disambiguates them through self-guided exploration, next assimilates targeted human corrections that handle out-of-distribution variations, and finally recovers the abstracted-away details online through compliant interaction. Because the resulting policy follows constraints rather than mimics trajectories, it generalizes across object poses, local geometries, and unmodeled contact dynamics. We validate our approach on seven real-world multi-stage contact-rich manipulation tasks and achieve over 90% success. These extensive experimental results establish environmental constraints as fundamental building blocks for efficient generalization in learning from demonstration.

Xing Li, Oliver Brock

Robotics and Biology Laboratory, Technische Universität Berlin · Technische Universität Berlin · Science of Intelligence, Research Cluster of Excellence, Berlin +1

Robot Skill LearningContact-Rich Robotic ManipulationGeneralization in Robotic ManipulationImitation LearningRobotic ManipulationLearning from Demonstration+1
AskPDF
Jul 15, 2026

Learning Forward & Reverse Skills from a Single Unfinished Demonstration for Constrained Manipulation Tasks

Learning from demonstration (LfD) enables robots to learn manipulation skills directly from expert demonstrations but remains challenging for contact-rich tasks involving geometric constraints and force interaction. Existing approaches typically require multiple complete demonstrations and do not support reverse skill execution. In this paper, we present a unified one-shot framework for constrained manipulation that learns both forward and reverse execution from a single, possibly unfinished demonstration. Our method decomposes demonstrations into non-contact and contact phases, with non-contact motion encoded with dynamic movement primitives (DMP), and contact motion represented as a sequence of screw motion primitives segmented by our proposed geometry-driven twist-direction segmentation algorithm. During execution, screw primitives are executed sequentially under admittance-guided pose correction and speed regulation, enabling task completion beyond the demonstrated trajectory length as well as reverse skill execution without additional learning data. Experiments on peg insertion, battery insertion, lock opening, and screw driving tasks demonstrate improved success rates and robustness over segmentation and one-shot trajectory learning baselines. Details are available on the project website: https://tuwien-asl.github.io/LfD-Screw/.

Yexin Hu, Haoyi Zheng, Johannes Heidersberger +1

Autonomous Systems Lab, Technische Universität Wien (TU Wien), Austria · Institute of Robotics and Mechatronics, German Aerospace Center (DLR), Germany

Robot Skill LearningContact-Rich Robotic ManipulationRobot Motion GenerationRobotic ManipulationLearning from Demonstration
AskPDF