1st workshop on iCARE @ CoRL2025

intelligent Cobodied Assistance and Robotic Empowerment

September 27, 2025 | Seoul, Korea

About iCARE Workshop

Welcome to the 1st workshop on iCARE @ CoRL2025: Intelligent Cobodied Assistance and Robotic Empowerment.

Worldwide, over 2.5 billion people need one or more assistive tools (e.g., wheelchairs, hearing aids, communication devices, seeing aids) and nearly 1 billion have no access to them. Assistive technology, involving the integration of robotics, AI, and human-centered design to empower people with diverse abilities, is fostering interdisciplinary collaboration.

The accessible and inclusive robot learning technology offers a truly "life-changing" potential. By providing intelligent cobodied assistance – where robots work intuitively with humans – we can open crucial doors to employment, education, and independent living for individuals with disabilities and the elderly.

From aiding to empowering, the workshop explores robotic empowerment, such as surgical assistive robots, exoskeleton, AR/VR glasses, embodied/cobodied AI, multilingual agents, driving aids, those that can help normal end users and improve their abilities. Indeed, robots possess "the potential to empower persons with diverse (dis)abilities".

The 1st iCARE workshop @CoRL2025 will provide a half-day session (poster, oral, invited talks) to integrate insights from assistive technology and cobodied AI into the robotics and learning community. We believe attendees can benefit from this interdisciplinary workshop.

📚 List of Topics

Topics of interest include but are not limited to the following:

VLMs for accessibility: How can LLMs and VLMs be used to describe environments or objects to disable users, or to interpret user commands?
Policy learning for assistive tasks: How can reinforcement learning (RL) enable assistive or wearable robots to empower the end users safely and adaptively?
Lifelong and few-shot learning: How can robots rapidly adapt to new users or environments with limited data or without access to the training data or model weights?
Simulation-to-Real transfer: How to use high-fidelity simulators to pretrain assistive behaviors before real-world deployment?
Efficiency and portability: How efficient models can be developed and deployed in edge devices that have limited computational resources?
Wearables and sensors: How can wearable devices (smart canes, vision aids, haptic feedback, etc) be integrated with robots to improve perception, planing, and performing?
Multimodal perception and personalization: How can robots fuse vision, touch, speech and contextual cues to understand user needs, and how can models be personalized to individual disabilities?
Inclusive mobility and navigation: What novel sensors, planning methods, or end-to-end learning approaches can give autonomy to wheelchairs or navigation aids for the visually impaired?
Benchmarking and datasets: What kinds of new datasets are necessary and generated by involving the end users during the data generation process, like collection, annotation, evaluation?
Formative and user study: How to design studies in different phases of system development, by involving the end users?
Human-robot interaction: How the assistance and robots are easy-to-learn, easy-to-use, and long-term usable with friendly user experience?

📢 Call for Papers

Authors are invited to submit papers within the scope of assistive intelligence (submission deadline: TBD). The workshop topics include (but are not limited to):

VLMs for accessibility: How can LLMs and VLMs be used to describe environments or objects to disable users, or to interpret user commands?
Policy learning for assistive tasks: How can reinforcement learning (RL) enable assistive or wearable robots to empower the end users safely and adaptively?
Lifelong and few-shot learning: How can robots rapidly adapt to new users or environments with limited data or without access to the training data or model weights?
Simulation-to-Real transfer: How to use high-fidelity simulators to pretrain assistive behaviors before real-world deployment?
Efficiency and portability: How efficient models can be developed and deployed in edge devices that have limited computational resources?
Wearables and sensors: How can wearable devices (smart canes, vision aids, haptic feedback, etc) be integrated with robots to improve perception, planing, and performing?
Multimodal perception and personalization: How can robots fuse vision, touch, speech and contextual cues to understand user needs, and how can models be personalized to individual disabilities?
Inclusive mobility and navigation: What novel sensors, planning methods, or end-to-end learning approaches can give autonomy to wheelchairs or navigation aids for the visually impaired?
Benchmarking and datasets: What kinds of new datasets are necessary and generated by involving the end users during the data generation process, like collection, annotation, evaluation?
Formative and user study: How to design studies in different phases of system development, by involving the end users?
Human-robot interaction: How the assistance and robots are easy-to-learn, easy-to-use, and long-term usable with friendly user experience?

📤 Submissions

Following instructions of the main conference CoRL:

Anonymization (Double-Blind Review): All papers must be anonymized for the double-blind review process.
Post-Acceptance De-anonymization: Accepted papers and reviews will be publicly accessible and de-anonymized after decisions are announced.
Reviewer Allocation: The aim is to have at least two reviewers per paper.
Publication Venue: Accepted papers will appear in the Proceedings of Machine Learning Research (PMLR).
Strict Deadlines: Abstract and paper submission deadlines are final, and no accommodations will be made for missing them.
Submission Platform: Papers must be submitted through OpenReview via this link: TBD
Page Limit: Up to 8 pages, excluding references, acknowledgments, and appendices
LaTeX Template: Use the provided LaTeX template: Download