A Community with Autonomous Personal Vehicles

1
A Community with Autonomous Personal Vehicles

• Larry Jackel
• larry.jackel_at_att.net

CyCab cybercar in La Rochelle (Robosoft,
France))
2
DARPA Challenge EventsGrand Challenge 2004 150
miles mostly dirt road No winners best
performance CMU 7 milesprimarily required good
GPS waypoint followingGrand Challenge 2005
132 miles Stanford Stanley winner, CMU
2ndMostly waypoint followingUrban Challenge
200760 miles in vacant military baseCMU lead
winner, Stanford 2ndTraffic rules and slow
moving trafficGPS navigation, some obstacle
avoidance
3
Why now? There is a convergence of forces

• Traffic jams are too costly in fuel and time
• We are running out of space to build new highways
  we need to make more efficient use of existing
  roadways
• Recent events shows the needed technology is
  within reach
• DARPA Urban Challenge

Urban Challenge National Qualifying Event(video
Not so easy
4
Why cant we just roll out APVs?

• The technology is still immature
• A protected environment is required
• Fixed infrastructure can greatly simplify the
  navigation and obstacle-avoidance tasks
• A community planned from the ground up for
  APVs can jump-start deployment. As the
  technology matures APVs can be released into the
  larger world.

5
How infrastructure can help

• In the Urban Challenge all vehicles were
  independent
• All sensors were on-board
• No vehicle-to-vehicle comms
• No guarantees of accuracy of route data provided
  by DARPA
• Deliberate blockages
• In our community we can have centralized Vehicle
  Traffic Control (VTC)
• No traffic jams, no ambiguity at intersections
• Fixed sensors can alert vehicles to obstacles and
  can augment on-board sensors
• Easy for fixed sensors to detect changes and
  locate obstacles
• As a backup, a network of cameras can let a human
  controller decide if a detected obstacle is
  actually a false alarm

6
Benefits beyond just APVs

• New urban architecture
• No more half-plane parking lot strip malls
• Grandma is mobile at night
• Moms Taxi doesnt need Mom
• Automated delivery of goods
• Completes the distribution chain for a robotic
  warehouse
• Use vehicles that are right sized for the job,
  no life support for driver required.
• More opportunities for robotics at each end

Replace with offsite robo self valet parking
7
Why cant we just roll out self-driven vehicles?

• The technology is still immature
• Cant drive in heavy traffic
• Cant drive with required safety margins
• Cant see people, dogs, holes in the road, etc
• But
• Comms and infrastructure cangreatly simplify the
  navigation and obstacle-avoidance tasks

DARPA knew the position off all vehicles during
the event and could have performed traffic
control and guidance
8
Marine Automated Information System
9
How infrastructure can help

• In the Urban Challenge all vehicles were
  independent
• All sensors were on-board
• No vehicle-to-vehicle comms
• No guarantees of accuracy of route data provided
  by DARPA
• Deliberate blockages
• Fixed sensors can alert vehicles to obstacles and
  can augment on-board sensors
• Easy for fixed sensors to detect changes and
  locate obstacles
• As a backup, a network of cameras can let a human
  controller decide if a detected obstacle is
  actually a false alarm

10
Next Steps for Research

• Detect moving obstacles in unstructured
  environments (animals, other vehicles)
• Detect people (moving and non-moving) in all
  poses
• Scalable Vehicle Traffic Control Model
• Vehicle-to-vehicle comms and vehicle-to-infrastru
  cture comms
• with service guarantees in heavy traffic
• Need Standards and APIs NOW so that the fleet
  will be ready when the technology arrives

11
EU CityMobil Project

• 2006 2011, 28 partners, 40M Euro budget
• 3 deployment sites
• Personal Rapid Transit at Heathrow (dedicated
  guideways)
• Automated shuttles (cybercars)fromfrom parking
  lots to trainstation and Rome expo center
• Dual mode buses in Spanish town of Castellon on
  20km corridor
• Vehicles appear to have dedicated roadways, or
  will move at low speed while autonomous
• Far more ambitious than any program in US
• Emphasis in US is on independent autonomy with
  gradual increase in inter-vehicle comms minimal
  dependence on infrastructure