{
  "abstract": "In modern robotics, learning-based policies are increasingly paired with model-based controllers to produce behaviors that are both adaptive and physically feasible. Yet these controllers -which enforce constraints such as kinematic feasibility, joint limits, collision avoidance or balance -are usually treated as non-differentiable black boxes: no gradient flows back through them to the policy learning pipeline. This disconnect prevents gradient-based policy learning from accounting for the physical constraints that actually shape the executed motion: the policy is trained to describe a trajectory rather than to produce one, and compensating for this typically requires large datasets or hand-tuned heuristics. We address this gap by introducing IKLayer, a fully differentiable wholebody layer based on differential inverse kinematics. IKLayer solves a cascade of parametric quadratic programs that encode physical constraints and differentiates through each program via implicit differentiation. To bridge the Euclidean output of neural networks and the SE(3) input poses required by the IK controller, we adopt a differentiable parameterization R 9 → SE(3) and derive its analytic Jacobian, which is free of periodic singularities and produces well-conditioned gradients uniformly over SE(3). On a simulated humanoid reaching task where balance constraints make a single-arm motion infeasible, training through IKLayer discovers a counterweight strategy in minutes, whereas zero-order baselines fail to solve the task. More broadly, these results suggest that exposing a controller's constraints to the policy's gradients turns constraint satisfaction from something the policy must learn into a property of the architecture.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": null,
  "author_total": 4,
  "authors": [
    {
      "name": "Mathis Scheffler"
    },
    {
      "name": "Yann de Mont-Marin"
    },
    {
      "name": "Cordelia Schmid"
    },
    {
      "name": "Justin Carpentier"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/bm6st14ksp9na.bib",
  "cited_by": 0,
  "doi": null,
  "fulltext_url": null,
  "id": "bm6st14ksp9na",
  "issns": [],
  "landing_url": "https://hal.science/hal-05778276v1",
  "language": "en",
  "licence": "CC-BY-NC-4.0",
  "licence_label": "CC BY-NC 4.0",
  "licence_source": "article",
  "licence_url": "https://creativecommons.org/licenses/by-nc/4.0/",
  "metadata_licence": null,
  "native_subjects": [
    [
      "hal",
      "0.info"
    ],
    [
      "hal",
      "0.stat"
    ],
    [
      "hal",
      "1.info.info-rb"
    ],
    [
      "hal",
      "1.stat.ml"
    ]
  ],
  "openalex": null,
  "pdf_url": "https://hal.science/hal-05778276/document",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-05",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/bm6st14ksp9na.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:17:47Z",
      "native_id": "hal-05778276",
      "source": "hal",
      "url": "https://hal.science/hal-05778276v1"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "hal:0.stat",
      "subfield": "statistics"
    }
  ],
  "title": "IKLayer: A Differentiable Inverse Kinematics Layer for Whole-Body Control",
  "title_text": "IKLayer: A Differentiable Inverse Kinematics Layer for Whole-Body Control",
  "type": "preprint",
  "updated_at": "2026-10-30T12:17:47Z",
  "url": "https://research.draconitic.com/paper/bm6st14ksp9na",
  "venue": "HAL",
  "versions": []
}