Early driverless vehicles: remote commands were not autonomous navigation

Removing a person from a vehicle's cabin does not, by itself, show that the vehicle can decide where to go. Early remote-control demonstrations and later autonomous-driving competitions solved different problems. That distinction is more useful than a single list of supposedly first driverless cars.

The February 23 Ars Technica history linked Leonardo Torres Quevedo's Telekino with later experiments in radio control, road-guided vehicles and onboard autonomy. Its broad historical arc is a starting point, rather than evidence that all those machines used the same architecture or demonstrated the same capability.

The earlier UnliveNews account extended the timeline with unsupported milestones and present-day robotaxi generalizations. This corrected version focuses on two documented stages: commands sent from elsewhere and a vehicle navigating a defined course without a person issuing each driving command.

Telekino kept the command source outside the vehicle

A March 2007 account by Spain's telecommunications historical forum describes the Telekino work recognized by an IEEE Milestone. It traces the idea to testing airships without exposing a pilot to the associated danger, using a transmitter on the ground and a receiver aboard the craft.

The account dates the first French patent application to December 10, 1902. It describes practical trials with a tricycle in Madrid's Beti-Jai building, followed by boat experiments and demonstrations in Bilbao in November 1905 and September 1906.

That sequence documents remote execution of movement commands. It does not establish independent perception of traffic or route planning. The forum's account also describes institutional funding for experimental work before later commercialization difficulties, making a blanket story of no official support too simple.

The historical distinction is not a criticism of the achievement. Separating an operator from the machine being controlled was a substantial technical step even though the operator remained the source of instructions.

A public demonstration and a navigation test ask different questions

The Ars feature describes radio-controlled road spectacles and experiments in which infrastructure guided a vehicle. These examples show why the label driverless can hide important differences. A human can be absent from the driver's seat while still providing the decisions, or guidance can depend on a specially equipped road.

Comparing such systems requires identifying what the demonstration actually removed: the driver's physical presence, the mechanical connection to controls, the need to steer continuously or the dependence on external guidance. Treating those as one capability makes the history look smoother than the engineering evidence warrants.

This account does not assign every historical first or verify an exhaustive chronology. It explains why a particular demonstration's conditions are part of its result rather than incidental scenery.

DARPA's desert challenge made course completion explicit

DARPA's Grand Challenge timeline records that no entrant completed the March 13, 2004 route from Barstow, California, to Primm, Nevada. The million-dollar prize went unclaimed.

For the October 8, 2005 event, the agency records five course finishers and a Stanford victory over a 132-mile desert route. Those outcomes describe performance on specified courses, not certification for unrestricted public-road operation.

DARPA's ten-year retrospective adds that the best 2004 vehicle traveled only 7.5 miles. It describes the 2007 Urban Challenge as a staged city course with moving traffic, obstacles and traffic-rule requirements; six teams completed it. Changing the test environment changed the task being demonstrated.

Historical progress does not erase operating conditions

These records support a history of increasingly demanding experiments without proving that every later deployment descends from one invention or that a competition result establishes universal driving ability. A remote-control trial and an autonomous navigation course can both matter while remaining different accomplishments.

The recovered article removes predictions about future regulation, fleet economics, all-weather capability and universal remote-assistance practices. It keeps the narrower lesson: the absence of a visible driver is the beginning of the question. The more informative evidence identifies who supplies decisions, what guidance the vehicle uses and which environment it has actually demonstrated it can handle.