Deadline Date: June 18, 2026
The National Research Council (NRC) is seeking the development of an integrated virtual and physical training system that combines human-in-the-loop control and machine learning to enable semi-autonomous robotic manipulators to perform contact-based tasks in unstructured environments.
The opportunity focuses on integrating virtual simulation and physical robotic operation, human-in-the-loop control, machine learning for adaptive manipulation, contact-based task execution in unstructured environments, force-enabled operator interfaces, visual and kinesthetic/haptic feedback, online learning algorithms, virtual sensor integration, task prototyping and validation, and semi-autonomous manipulation for applications such as infrastructure inspection, maintenance and repair, search and rescue, medical interventions, and Explosive Ordnance Disposal (EOD) operations.
The proposed system is intended to address the current lack of integrated online learning capabilities that allow robotic systems to seamlessly transition between virtual simulation and physical operation while incorporating human inputs. Through a combination of pre-programmed behaviors and operator commands delivered via a force-enabled input device, a virtual robot is expected to interact dynamically with scenes and objects using a gripper or specialized tools. Operators will receive feedback through visual displays and haptic interfaces, enabling more effective task execution and training.
Virtual environments and objects may be created using computer-aided design (CAD) tools or generated from digitized real-world elements. The system should support the development and implementation of online machine learning algorithms by utilizing data from virtual sensors, including visual sensors, robot encoders, force sensors, and input devices. This learning framework is expected to facilitate continuous adaptation and performance improvement during robotic operations.
Following task prototyping in the virtual environment, learned behaviors and trajectories are to be further trained and validated using the same operator interface connected to a hardware manipulator operating in a physical environment. This arrangement enables the collection of force information and sensor feedback generated through real-world interaction, providing more realistic insights into system dynamics and contact conditions while supporting ongoing online learning during human interaction.
The proposed solution is expected to achieve a minimum Technical Readiness Level (TRL) of 5, demonstrating the capability to prototype and test the system within a laboratory environment. Although full field deployment is not required, applicants are encouraged to consider factors important for future operational use, including compactness, ruggedness, reliability, stiffness-to-weight ratio, and power requirements.
The challenge is offered under Phase 2 and provides funding of up to CAD 1,500,000 for projects with a duration of up to 18 months. An estimated one grant is expected to be awarded.
Eligible applicants must be for-profit businesses incorporated in Canada, either federally or provincially, with 499 or fewer full-time equivalent employees. Research and development activities must take place in Canada, and applicants must meet requirements related to Canadian workforce participation, compensation, employee location, and executive residency.
For more information, visit Government of Canada.
