Hair Cutting Robot (HCR)

Highlights

• Two editors: absolute servo angles, and a Cutter Grid language that compiles deterministic inverse kinematics against certified motion profiles
• Three executors run one frozen program IR: the browser preview, the Rust server replay that owns the official score, and no_std firmware shared with the arm
• A student gets a tutorial, thirteen lessons, solo practice, and competitive Versus rounds on a server clock. Every score stays hidden until the round closes.
• The arona CAT engine sets the difficulty. A 2PL IRT model re-estimates each learner from the server-replayed scores and selects the next challenge.
• The authoring tool rejects a generated challenge unless a reference solver can solve it, so no student receives an impossible task
• Collision handling is deterministic. The program stops at the last safe pose and names the Blockly block that caused the collision, so the student can repair the program.
• Playwright runs the acceptance tests in Chrome and Edge: end-to-end suites, visual suites, and Cutter Grid audit suites
• Wrote the system specification and the phased implementation plan that the student teams build against, and the InnoCarnival 2026 pavilion deck for the PolyU exhibition
• Built the base framework and the content for the two academic event sites of the project: JHM-SRAS 2026 (sras.hcr.rs) and RoboEdu@ICWL–SETE 2026 (robo.hcr.rs)
• Mentored ten student teams on system design, implementation, and technical writing
• The course has run for three cohorts. The largest is at the Indian Institute of Technology Kharagpur, and it is also the cohort that is furthest from the hardware.
• No physical arm travels to Kharagpur. The browser client and the server-side replay are the whole teaching apparatus. This is what makes a cohort of that size possible at a distance.
• The Kharagpur participants already know how to program, so the course spends its time on the arm and not on the syntax

Tags