1. Agile Robotics Program Review AR Team MIT, Lincoln Lab, Draper Lab, BAE August 8, 2008

2. Programmatics Budget snapshot Platform costs Timeline and milestones Year 2, 3 directions Summary and discussion

3. Budget Snapshot Notes: 1) Major fabricated and non-fabricated equipment purchases planned for period after July 2008 delivery of prototype vehicle 2) Significant travel and shipping costs anticipated in February and March 2009 to support end-of-year prototype capability demonstration

4. Drive-by-Wire Platform Costs Forklift capacity Manual forklift base cost Drive-by-Wire Conversion (parts, labor) Sensors, Power, Wiring Computers, PDA, Intent Annunciators Total System Cost Premium (percent of base cost) 3,000 lb.$25K$20K$35K$20K$100K300% 3,000 lb.*$50K$20K$35K$20K$125K150% 10,000 lb.$90K$25K$40K$25K$180K100% 17,500 lb.$125K $30K$45K$30K$230K84% 33,000 lb.$250K $35K$50K$40K$375K50% 45,000 lb.$300K$35K$50K$40K$425K42% 50,000 lb.$320K$40K$50K$40K$450K41% 80,000 lb.$450K$40K$60K$50K$600K33% Notes: *A 3,000 lb. Yale forklift costs $50K; we chose a US-manufactured Toyota forklift for $25K All costs are estimated; procurement lead time approx. 12 weeks Costs exclude robotic arm(s) under consideration for future years

5. Working Timeline Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr Y2 Year 1 Requirements analysis, system & interface design, safety, user testing Simulation development Mockup forklift Full-scale prototype Capability migration Visit to Visit to Visit to CASCOM Incursion Seamless Shouted Testing & LIA CASCOM Ft. Campbell Visit to MIT detection handoff warnings validation Port DGC Speech Gesture Pallet Situational Whole-SSA codebase integration support detection awareness simulation Drive-by-wire Sensor Planning Pallet prototyping placement and control engagement Drive-by-wire Sensor Planning Pallet Pallet Vehicle prototyping placement and control engagement mobility interaction Lincoln FMTV Laboratory studies Kickoff Meeting Program Review Year 1 Demo (Rented forklift)(Purchased forklift) Program Review Today Preliminaries: Drive-by- Wire Platform Development Prototype Autonomous Capability Development

6. Sept. 2008 Milestones ( = anticipated) Capability 0.0: 30 Jun 08 – Analyze and summarize existing SSA operational practices – Demonstrate low-fidelity simulation of terrain, pallets, and trucks – Develop prototype speech, gesture interface to command forklift Capability 0.1: 30 Sep 08 – Build a partially-actuated mockup forklift – Develop prototype sensing, planning, control algorithms – Demonstrate mockup working in a real-world scenario to identify, localize, select, engage, lift, transport, and place one pallet at a time as directed Capability 0.2: 31 Dec 08 – Convert a stock manual forklift to drive-by-wire control – Team members complete required OSHA 1910.178 training – Understand forklift degrees of freedom, sensing, feedback and control –Develop functional interface, perception, planning, control algorithms Interface: natural interaction, effective for cluttered real-world situations Perception: multiple trucks and pallets, uneven terrain, other moving agents Planning: summoning; pallet approach, engagement, transport and placement Control: fast, accurate mobility and manipulation; robustness to uneven terrain Capability 0.3: 31 Mar 09 –Prototype demonstration at venue to be determined (tentatively Ft. Belvoir)

7. Planned Year 2, 3 Directions Higher-level reasoning –E.g. moving blocking pallets aside to achieve goal –With supervisor direction, or fully autonomously –Higher-level direction (e.g., “put those pallets onto…”) Fine-grain (sub-pallet) manipulation –Robot arms, pallet break- down, box manipulation –Integration with RFID, NSN, parts catalogs Adverse environments –Mud, rain, snow, dust, sand-storms; at night –GPS-denied conditions (e.g., landmark-based navigation) Many forklifts, few supervisors –Human dispatcher, whole-SSA interface –Multi-task optimization Passive (video) sensing –Monocular or stereo –More CPU resources –Cheaper, less emission Year 1 Capability Whole-SSA transfer –Roll up onto many trucks –Roll out SSA as specified Primary focus areas

8. Summary Tackling problem along four fronts, in parallel: –Requirements analysis: LIA, CASCOM, Ft. Campbell –Simulation studies for sensing, dataflow, throughput –Mockup experiments, drive-by-wire elements –Full-scale prototype development, data collection At present, substantially ahead of schedule –Several September, December milestones achieved –Demonstrated components in simulation, on mockup –(Partially) working full-scale prototype by fall 2008 Anticipated fruitful directions for Years 2, 3 –Improved manipulation, reasoning, sensing, scaling

9. Discussion

10. Adjournment of MIT Team

11. Program Management Discussion 0900 – 0930: Arrive MIT Kiva conference room (32G-449) 0930 – 0945: Informal introductions 0945 – 1030: Summary goals and status 1030 – 1045: Break, walk to demonstration venues 1045 – 1145: Demonstrations (Kiva, Holodeck, Hangar) 1200 – 1300: Lunch [Highlights of other MIT robotics] 1300 – 1415: Technical briefings (5 x 15 minutes) 1415 – 1430: Break 1430 – 1530: Technical briefings (4 x 15 minutes) 1530 – 1630: Feedback and discussion 1630: Main group adjourns 1630 – 1700: Program management discussion 1700: Program review adjourns

12. Program Management Discussion

13. Program Review Adjourns Travel safely!