Former Employees Blame Erratic Musk & Costing For FSD Technology To Not Be Viable

For quite some time now, Elon Musk has been determined to make Tesla's cars capable of self-driving. However, a report in the Washington Post points out that this ambition has been hindered by a combination of aggressive cost-cutting measures and inconsistent management, a fact that has been highlighted by his handling of the Twitter acquisition.

Because of the global semiconductor crisis, the billionaire decided to reduce costs and simplify the supply chain. To do this, he removed the radar sensors that were used to detect hazards from a long distance and prevent cars from crashing into them. He argued that Teslas already had eight cameras to see what's on the road and spot hazards in each direction. This was the foundation of the AutoPilot ADAS feature that worked with the removed radars.

According to Musk, cameras alone were enough, but some Tesla engineers disagreed. They even asked former executives for advice on how to persuade Musk to keep radars in the system, but they were unsuccessful. Musk removed radars from cars that had already been shipped with them and stopped installing radars in new vehicles.

Shortly after the radars were removed, problems began to occur. Without the radar, Teslas became more vulnerable to simple perception errors when the cameras were obstructed by raindrops and even bright sunlight.

According to former employees, safety officials, and other experts, Tesla's cars were involved in more crashes and near misses and making more embarrassing mistakes. These issues were caused by a lack of critical equipment.

When Musk talked about Full Self-Driving, which came after AutoPilot, he said it was the most important thing that would make Tesla worth a lot of money instead of worthless.

However, Tesla's vision is now having problems. The technology has been recalled and its rollout has been suspended because officials in the US are worried that the cars won't follow speed limits or stop at stop signs.

Many customers have sued Musk for saying that Tesla's technology is better than it actually is. The US government and regulators are also checking Tesla's system because it has many safety problems.

Previous employees say that Tesla's quick pace of work, cost-cutting measures, and Musk's insistence on non-traditional methods have caused unique problems at the company. His leadership style has also contributed to issues as he has pushed his team to deliver features before they were fully tested for public use. There are concerns that Tesla's software may not be safe to use on public roads.

John Bernal, a former Tesla test operator who worked on AutoPilot, said that the development of the system was slow inside the company. He was fired for misusing the technology when he posted videos of Full Self Driving in action.

"Elon keeps tweeting, 'Oh we're almost there, we're almost there,'" Bernal said. However, "internally we're nowhere close," so now we have to work harder and harder to reach our goals.

Furthermore, since Musk acquired Twitter, he has transferred numerous engineers from the AutoPilot division to work on code at Twitter.

According to Bernal, no one realized what it was like to work for Elon until they saw how he managed Twitter. Bernal described Twitter as "just the tip of the iceberg" in revealing how Elon operates Tesla.

During Tesla's Autonomy Investor Day in 2019, Musk announced that more than one million Tesla cars equipped with full self-driving hardware would be on the road by 2020. This claim caused Tesla's stock price to soar, making it the most valuable automaker in the world and propelling Musk to the top of the list of the world's richest people.

Musk had gathered a great team of engineers who were willing to work hard on the problem. Musk himself, along with other executives, tested the FSD software and also gathered requests for their engineers to fix problems.

However, there was no clear plan in place, while Waymo, a pioneer in self-driving technology owned by Alphabet, had strict testing protocols. Eventually, Tesla released the FSD (Full Self Driving) update to 360,000 owners who paid $15,000 to be eligible for the feature and allowed them to activate it at their own discretion.

Tesla, Waymo, and even Apple had different approaches to solving the problem. Tesla's philosophy was to gather more data by having users use the software in the real world. This allowed rapid data collection and with the help of neural networks, the software was able to learn on its own.

Tesla's competitors were cautious and created rules to address any issues with the self-driving software. They also used LiDAR and radar sensors, which helped the software better understand its surroundings.

Tesla's system failed because it did not have some important sensors. Tesla is also going to remove other sensors that detect obstructions that are very close to the car. This could make Tesla's system even worse.

According to Brad Templeton, an expert in self-driving cars who worked on Google's self-driving car, LiDAR has a big advantage. It can always detect a train or truck, even if it doesn't know what it is. The technology can tell that there's an object ahead and the car can stop without needing to know anything else.

After radars were removed, there were more phantom braking issues. There were 107 complaints over three months before, and 34 complaints over 22 months after. Within two weeks, the NHTSA received 250 complaints about the issue. Because a dozen Teslas crashed into parked emergency vehicles, the NHTSA investigated AutoPilot.

"It's not the only reason they're experiencing difficulties, but it's a significant factor," stated Missy Cummings, a former senior safety advisor for NHTSA who has criticized the company's approach and recused herself from matters related to Tesla. "The radar assisted in detecting objects in the forward field. As computer vision is susceptible to errors, it provides a sensor fusion method to verify if there is an issue," she further explained.

According to a former employee, Musk's unwillingness to listen to suggestions created an atmosphere of fear at Tesla. Engineers who disagreed with him were fired. Since many Tesla employees also work at Twitter, progress in software updates for AutoPilot and FSD beta has been slower than expected.

Ross Gerber, a Tesla investor who hopes to join the Tesla board, wants Musk to work full-time at the car maker. Meanwhile, Tesla has lost Andrej Karpathy, its director of artificial intelligence, who went on a sabbatical and then joined OpenAI, the company behind ChatGPT.

Reportedly, Ashoke Elluswamy, who was the director of Tesla's AutoPilot, is now working at Twitter. Additionally, an anonymous Tesla employee has left to work for Waymo.