3. Recent Work Creating a useable interface New methods for row detection New methods for turning and row entry
4. Autonomous Vehicle User Interface Goal: Create an interface for the vehicle useable by growers and workers without engineer assistance Approach: Follow structured, formal human-computer interaction design methodologies Interview workers to understand farm operations and equipment Define requirements that must be satisfied in the final interface product Follow a top-down design approach from requirements to functionalities Test with the workers on the APM
6. Existing Interface It tells us a lot Good for testing and debugging Hard to use even for us Impossible to use without lots of training Scary for those without computer backgrounds
7. Designing a New Interface Start from the beginning What does the vehicle have to do? Who will be using it? What is their current experience/expectation?
19. User Tests: First Round Interface run on laptop, controlling simulated vehicle Workers completely unfamiliar with interface Presented with similar scenarios as before Told to use vehicle in a variety of tasks
20. First Round Results Computer-experienced users felt comfortable Others were afraid to give an incorrect command Autonomous vehicle is an expensive piece of equipment Conclusion: Users need some training, even if it is short
21. Final User Tests Interface stayed the same Went to Hollabaugh Orchards, Biglerville Four users operated APM Owner/manager and three workers Two workers understood some English,one understood very little
22. Final User Tests All were able to control APM All understood speed, offset, and go/stop mode Many mentioned potential applications for which they would like to use the APM One common complaint Need for speed control in stop/go mode http://www.cs.cmu.edu/~casc/public/showcase_video1.wmv
25. Conclusions This interface can make APMs easy to use by: Reducing text on screen Presenting vehicle’s location clearly Abstracting orchard tasks to single button clicks
26. Conclusions In order to make performing tasks simpler for workers,additional setup is required by managers
28. Recent Work Creating a useable interface New methods for row detection New methods for turning and row entry
29. Previous Row Detection Susceptible to spurious data Apply some filters to reduce the effect of bad detections But we want to have no bad detections
30. New Row Detection Work Using particle filter to track row over time Looks good so far Needs more testing
31. New Row Entry Work Make wider turns Points vehicle at row sooner Easier to detect row Smoother turns in tight spaces Account for trailer when turning Use high accuracy localization (Area 1.2 work) to improve vehicle control
32. Back to the Orchards May 24: N. Blosi platform testing in Biglerville, PA June 7: APM testing in Biglerville July 19: APM testing in Washington
36. Augmented Harvest DBR Conveyor Concepts update Statement of work and contract finalized Work initiated New bin filler head looks good Laying out power unit with integrated exhaust silencing and intake filtering Ordering parts for hoses and vacuum chambers Bubble pack bin filler Robin Pritz and Gwendolyn Barr completed first bruise testing Method shows promise, but new prototype is required for field testing
43. Summary Bubble pack bin filler shows promise Bruising is not bad given distance apples are tossed Upcoming improvements More guiding by net Better distribution of bubble pack layers Lighter and better padded frame Mechanism for raising and lowering frame Work with DBR Conveyor Concepts is ramping up