# PWR-214 — Shared autonomy: learner worksheet Release: Revision 7T · Full Tutorial Edition Estimated time: Estimated 14 min reading; practical time is provider-set Difficulty: Intermediate Equipment: Specialist equipment — Declared Shared autonomy fixture: A wheelchair-style simulator travels a marked virtual corridor. The person commands direction; automation may slow or steer around foam obstacles. A bright mode label shows MANUAL, AUTO or SHARED.; Setup aid for Make mode visible: Verify decision allocation, mode indicator, limits, override, logs and fallback.; Shared autonomy log: Task success, intent errors, override latency, workload, collisions and failure recovery versus manual, autonomous and shared control; retain Shared autonomy errors, assistance, stop and fallback.; Shared-autonomy evidence grid: use “Shared control of a medical robot with haptic guidance” and the BCI mobile-robot shared-control study to record person command, automation contribution, mode display, disagreement, override and outcome. Use the NIST Privacy Framework to define who may inspect command traces. The supplied corridor remains a simulator, and the qualified team controls mode changes and degraded-assistance tests. Space: Specialist setting Automated structural checklist: 10 of 10 structural checks present Evidence quality/context: G4; Deep research depth Automated method-quality band: Comprehensive Method-rating basis: Structure coverage, instruction/check specificity, alternative distinctness, source troubleshooting, comparison depth and whether purpose/check fields are authored rather than derived. Editorial review: Pending manual editorial sign off ## Before you begin - [ ] I read the tutorial authority and stop conditions. - [ ] I have the required equipment/space or a declared accessible alternative. ## Canonical source context The corridor simulator displays MANUAL, AUTO or SHARED while the person commands direction and automation may slow or steer around foam obstacles. Compare all three modes, record intent conflicts and override delay, then practise assistance loss. Safe handling here does not establish shared control of a real mobility or manipulation device. Target outcome: With a qualified team, the learner completes a simulated navigation or manipulation task in manual, automatic and shared modes, logs intent conflicts and demonstrates override. ### Authority and setup conditions - Obtain qualified clinical-engineering supervision and participant consent for the exact simulator or approved device. - Verify decision allocation, mode indicator, limits, override, logs and fallback. - Start check for Shared autonomy: Every decision has a human or machine owner. - Top-of-sheet stop for Shared autonomy: Stop for mode uncertainty, unexpected intervention, repeated intent error, failed override, lost link or limit breach. ### Required or supplied materials - Declared Shared autonomy fixture: A wheelchair-style simulator travels a marked virtual corridor. The person commands direction; automation may slow or steer around foam obstacles. A bright mode label shows MANUAL, AUTO or SHARED. - Setup aid for Make mode visible: Verify decision allocation, mode indicator, limits, override, logs and fallback. - Shared autonomy log: Task success, intent errors, override latency, workload, collisions and failure recovery versus manual, autonomous and shared control; retain Shared autonomy errors, assistance, stop and fallback. - Shared-autonomy evidence grid: use “Shared control of a medical robot with haptic guidance” and the BCI mobile-robot shared-control study to record person command, automation contribution, mode display, disagreement, override and outcome. Use the NIST Privacy Framework to define who may inspect command traces. The supplied corridor remains a simulator, and the qualified team controls mode changes and degraded-assistance tests. ## 1. Allocate decisions explicitly The team writes that the person chooses destination and may stop; automation may slow and propose obstacle avoidance within the corridor. Why: Every decision has a human or machine owner. Success check: Every decision has a human or machine owner. Accessible alternative: Ask the qualified provider to adapt positioning, cue modality, pace, interface or rest while preserving this step’s declared check: “Every decision has a human or machine owner.” If an adaptation changes the measured task or configuration, record it as a different configuration rather than an equivalent attempt. Alternative relationship: Conditional equivalence requires target and configuration check Learner notes: ________________________________________________________________________________ Completed: [ ] ## 2. Make mode visible Check the persistent mode label, transition sound or vibration and current override control. Why: The learner can identify the mode without guessing from motion. Success check: The learner can identify the mode without guessing from motion. Accessible alternative: Ask the qualified provider to adapt positioning, cue modality, pace, interface or rest while preserving this step’s declared check: “The learner can identify the mode without guessing from motion.” If an adaptation changes the measured task or configuration, record it as a different configuration rather than an equivalent attempt. Alternative relationship: Conditional equivalence requires target and configuration check Learner notes: ________________________________________________________________________________ Completed: [ ] ## 3. Set bounds and fallback Engineers configure speed, corridor, obstacle clearance, stop, manual/degraded mode and safe state. Why: The simulator cannot leave the declared envelope. Success check: The simulator cannot leave the declared envelope. Accessible alternative: Ask the qualified provider to adapt positioning, cue modality, pace, interface or rest while preserving this step’s declared check: “The simulator cannot leave the declared envelope.” If an adaptation changes the measured task or configuration, record it as a different configuration rather than an equivalent attempt. Alternative relationship: Conditional equivalence requires target and configuration check Learner notes: ________________________________________________________________________________ Completed: [ ] ## 4. Record manual baseline Complete the corridor in manual mode and log time, path error, interventions, workload and collisions. Why: Shared performance has a matched comparator. Success check: Shared performance has a matched comparator. Accessible alternative: Ask the qualified provider to adapt positioning, cue modality, pace, interface or rest while preserving this step’s declared check: “Shared performance has a matched comparator.” If an adaptation changes the measured task or configuration, record it as a different configuration rather than an equivalent attempt. Alternative relationship: Conditional equivalence requires target and configuration check Learner notes: ________________________________________________________________________________ Completed: [ ] ## 5. Run shared mode The learner states destination, automation proposes path changes and the learner confirms or overrides according to the protocol. Why: Each machine intervention and human response is logged. Success check: Each machine intervention and human response is logged. Accessible alternative: Ask the qualified provider to adapt positioning, cue modality, pace, interface or rest while preserving this step’s declared check: “Each machine intervention and human response is logged.” If an adaptation changes the measured task or configuration, record it as a different configuration rather than an equivalent attempt. Alternative relationship: Conditional equivalence requires target and configuration check Learner notes: ________________________________________________________________________________ Completed: [ ] ## 6. Name disagreements When automation steers right but the learner intends left, say “conflict,” stop or override and record whose evidence controlled the result. Why: Intent error is not hidden inside successful arrival. Success check: Intent error is not hidden inside successful arrival. Accessible alternative: Ask the qualified provider to adapt positioning, cue modality, pace, interface or rest while preserving this step’s declared check: “Intent error is not hidden inside successful arrival.” If an adaptation changes the measured task or configuration, record it as a different configuration rather than an equivalent attempt. Alternative relationship: Conditional equivalence requires target and configuration check Learner notes: ________________________________________________________________________________ Completed: [ ] ## 7. Practise degraded control The team removes automation assistance; the learner uses the approved manual or alternative fallback or stops in safe state. Why: Loss of automation does not cause continued uncontrolled movement. Success check: Loss of automation does not cause continued uncontrolled movement. Accessible alternative: Ask the qualified provider to adapt positioning, cue modality, pace, interface or rest while preserving this step’s declared check: “Loss of automation does not cause continued uncontrolled movement.” If an adaptation changes the measured task or configuration, record it as a different configuration rather than an equivalent attempt. Alternative relationship: Conditional equivalence requires target and configuration check Learner notes: ________________________________________________________________________________ Completed: [ ] ## 8. Compare all modes Review task success, intent errors, override latency, workload, collisions and agency in manual, auto and shared conditions; participant chooses next step. Why: Shared mode is not declared best from completion alone. Success check: Shared mode is not declared best from completion alone. Accessible alternative: Ask the qualified provider to adapt positioning, cue modality, pace, interface or rest while preserving this step’s declared check: “Shared mode is not declared best from completion alone.” If an adaptation changes the measured task or configuration, record it as a different configuration rather than an equivalent attempt. Alternative relationship: Conditional equivalence requires target and configuration check Learner notes: ________________________________________________________________________________ Completed: [ ] ## Reflection What changed? ________________________________________________________________________________ What remains difficult? ________________________________________________________________________________ What will I repeat, adapt, ask for help with, or stop? ________________________________________________________________________________ --- Completion of this worksheet demonstrates tutorial participation only. It does not establish capability, qualification, safety clearance, diagnosis, treatment or independent validation.