top of page
  • Writer's pictureFlow Australia

Tesla and Hyundai reveal plans to replace taxi drivers

Tesla CEO Elon Musk has promised to reveal a self-driving robotaxi in August but transport experts say autonomous technology is not ready for the road.



Self-driving cars do not get drunk, they do not fall asleep, they do not get distracted by text messages, and experts and manufacturers agree they could be the answer to slashing the road toll.


It's one of the reasons why autonomous vehicles are in the spotlight again, with Tesla promising to unveil a robotaxi in August and Hyundai showing off the results of its driverless car trial in Las Vegas. 


But debate is raging in the industry over whether the technology is or will ever be ready to drive in busy, unpredictable environments without any human oversight. 


Some transport experts say the technology will require a major breakthrough in artificial intelligence to achieve its potential, while others argue more data and trials could make the difference. 


On-road testing in the United States is failing to allay concerns, however, after a spate of breakdowns and unexpected accidents. 


Driverless vehicles sped back into the headlines last week when Tesla chief executive Elon Musk posted about them on X, promising a "Tesla robotaxi unveil on 8/8".


It's not the first time the company, renowned for its driver assistance technology, has raised the prospect of self-driving cars but the first time they have received a launch date.

Hyundai's self-driving 'robotaxi' completes a test in Seoul, South Korea.


Tesla's announcement came just days after Hyundai revealed its self-driving electric car, a modified Ioniq 5, had passed an examination similar to a US driver's test on roads around Las Vegas. 


A retired examiner rated the vehicle's ability to change lanes, perform manoeuvres, and obey traffic signals and signs.


Hyundai senior vice-president Sungwon Jee says the test was designed to demonstrate how far the technology had come and to promote trust in the vehicles. 


"We will continue our efforts to communicate our AI-based autonomous driving technology, which is set to provide positive changes to our customers' everyday lives," he said.


But two companies have already launched live self-driving car trials in the United States, with Google-based Waymo offering rides in parts of Phoenix, San Francisco and Los Angeles, and General Motors-owned Cruise operating in San Francisco, Austin, Phoenix and Houston.

Cruise had to suspend driverless services after one of its cars struck and dragged a pedestrian.


Cruise was forced to suspend all driverless services in October last year, however, after one of its cars struck and dragged a pedestrian for six metres. 


The company this week announced plans to start collecting traffic data again in Phoenix, Arizona, but only "using human-driven vehicles without autonomous system engaged".


Swinburne University of Technology future urban mobility professor Hussein Dia says the frightening incident was one of several accidents involving Cruise vehicles, which has created distrust and animosity towards the technology among some road users.


"One of the Cruise vehicles ... crews were fixing the road and the concrete had not set yet and the vehicle did not recognise this and it went straight through and got stuck in the concrete," he said.


"There's a lot of tension between people and these vehicles and, in my opinion, it's because these vehicles are not ready yet."


Other incidents during US trials have included near-misses with pedestrians, vehicles stopping in intersections without reason, and cars misjudging vehicles being towed or with unusual shapes, such as articulated buses. 


Despite these events, Prof Dia says the technology holds too much promise for all transport firms to give up on it. 


Years of further development, data collection, research and carefully monitored trials will be required, he says, to help the technology to reach its potential.


"We still need a major AI breakthrough," he says. 


"The robotaxis in the US are probably at level-four automation but to get to level five where the vehicle can go anywhere, any time, in any conditions, that needs much more development."


A recent survey of 562 Australians at Swinburne University found almost half liked the idea of self-driving cars, but nearly three in four respondents said the vehicles should still be able to be driven by a human. 


Dr Stewart Worrall, a senior research fellow with the University of Sydney's Australian Centre for Field Robotics, says preparing self-driving cars to react perfectly in all environments and scenarios was a huge undertaking, and one that would take years to tackle.


"There's always a trade-off because no system is perfect, like the current system of people driving is far from perfect," he said.


"The challenge is that autonomy is going to be held to a higher standard despite the fact that it might be as good or better than people."


Self-driving technology, he says, is currently best suited to areas such as "freeways and highways where there's not the complexity of being in a city," with pedestrians and unexpected driving behaviour, and improvements will require access to more traffic and trial data. 


Already, the Queensland Government and QUT have held trials of an automated vehicle called ZOE2 in regional towns including Mt Isa and Bundamba, autonomous shuttle buses have been tested in Sydney Olympic Park, and Transport NSW has established a Future Mobility Testing and Research Centre.

 

Dr Worrall says plans are also underway to set up a testing facility to focus on autonomous vehicles for use in rural and regional Australia, with July raised as a possible launch date. 


But the technology could still be 10 years away from widespread and reliable use, he says, and Tesla's August announcement may offer just a peek of the future.


"If I had to make a guess, I would say Tesla will reveal a new configuration of their car that is heading towards a future (robotaxi) when it's ready," he says.


"One of the things Tesla does have is a lot of data, though, and data is very important to solving these problems."


bottom of page