Autonomous driving technology, such as Tesla's Full Self-Driving system, has the potential to significantly improve road safety, but there are still challenges and limitations that need to be addressed before achieving completely autonomous driving on the roads
Questions to inspire discussion
-
What are the challenges of autonomous driving technology?
—The video discusses the challenges of GPS errors, sensor fusion, and navigating cluttered environments with autonomous vehicles.
-
What is the potential impact of Tesla's Full Self-Driving system on road safety?
—Tesla's Full Self-Driving system has the potential to significantly reduce fatalities and improve safety on the road, especially when compared to human drivers.
-
What is the speaker's experience with Tesla's Full Self-Driving feature?
—The speaker has extensive experience with autonomous vehicles, including driving and programming, and has purchased Tesla's Full Self-Driving feature for their Model Y.
-
How long will it take to achieve completely autonomous driving on the roads?
—The speaker suggests that it will take 15 to 20 years to achieve completely autonomous driving on the roads.
-
What is the goal of autonomous driving technology in the future?
—The goal is for vehicles to behave as a swarm for increased efficiency and intelligence, with the potential for widespread comfort and adoption of self-driving technology.
Key Insights
- 🚗 A personalized full self-driving solution may be necessary for each individual's specific needs and preferences.
- 🚗 Phase two of Tesla's AI training involves tuning it for different driving cultures and environments, like Boston versus Nebraska.
- 🚗 Maximizing the use of the road by packing cars closer together could eliminate the need for building more lanes and improve traffic flow.
- 🚗 The possibility of banning human drivers on certain roads due to safety concerns and the potential efficiency of autonomous cars communicating with each other is a fascinating concept.
- 🛣️ If every car had Tesla's full self-driving, the roads would be safer and autonomous driving would work better than it does right now.
- 🚗 The idea of a fake City with a thousand Teslas all full self-driving could be much safer than having humans and full self-driving in a mix.
- 🌟 "It's night and day better than what we had in 2008. It's way better, you know, I'm a huge Improvement."
- 🚗 The optimization function for autonomous driving includes safety, trip length, and passenger comfort, which can vary based on individual preferences and situations.
#FSDStories #DrKnow-it-all #DavidArmstrong
XMentions: @DrKnowItAll16
Clips
-
00:00 🚗 Dr. David Armstrong, an experienced autonomous driving engineer, reviews a video on autonomous driving technology and discusses his past projects and challenges in building unmanned systems for the military, comparing old robotics technology to the modern Tesla Model Y.
- Dr. David Armstrong, an experienced autonomous driving engineer, discusses his work in the 90s and reviews a video on autonomous driving technology.
- Computers were expensive and slow in the past, and the speakers both graduated high school in the early 80s and avoided using punch cards.
- The speaker discusses their experience in building unmanned systems for the military, including vehicles for range clearance and heavy construction equipment, as well as their involvement in the DARPA Grand Challenge Series.
- The speaker discusses the challenges and excitement of participating in autonomous vehicle competitions, including the difficulties of the route and the eventual success of Stanford's car.
- The speaker discusses their past projects and compares old robotics technology to the modern Tesla Model Y.
- The speaker discusses the visual detection system used in the test vehicle, including the VME back plane and GPS cost, and its application in computing the position and orientation of the vehicle.
-
06:57 🚗 The speaker reviews the challenges and limitations of autonomous driving technology, including GPS errors and the use of sensors and cameras in the Tesla full self-drive system.
- The speaker discusses the introduction of GPS technology and its impact on navigation, particularly in scuba diving, and how it has become a common tool for younger generations.
- Electrical engineers discovered that by analyzing carrier phase waveforms, they could achieve centimeter accuracy in driving paths with a robot, which was useful for pre-recording and generating sweep patterns for military missions.
- The speaker discusses the challenges of recording paths for autonomous vehicles in the desert and the unpredictability of the DARPA challenge.
- The speaker got stranded in the desert while trying to map out a location, had to call for rescue, and was relieved when they finally got out.
- The speaker discusses the challenges and limitations of autonomous driving technology, including a specific incident where a GPS error caused the vehicle to veer off course and get entangled in barbed wire.
- The speaker discusses the sensors and cameras used in the Tesla full self-drive system, comparing it to previous autonomous vehicle challenges.
-
17:43 🚗 The speaker reviews the technical details and future of autonomous driving, including military applications and the goal of vehicles behaving as a swarm for increased efficiency and intelligence.
- The speaker discusses the technical details of running a neural navigation system on a hybrid car, tapping into the car's power supply and dealing with the complexity of interfacing with the car's control systems.
- The speaker discusses the military applications of autonomous technology for sweeping and detecting unexploded ordnance, and mentions testing the technology on a university campus.
- The speaker went to Great Falls, Montana to sweep a farmer's field for Lewis and Clark's iron frame boat, but the archaeologists wouldn't let them dig it up.
- The future of autonomous driving involves multi-vehicle control and the goal is for vehicles to behave as a swarm for increased efficiency and intelligence.
- The speaker reminisces about using SGI machines in the 90s for graphic work and how slow and different the technology was compared to today.
- The speaker discusses the use of sonar sensors from Polaroid cameras and AAR search algorithms for path planning in military applications.
-
26:13 🚗 Tesla's Full Self-Driving system is limited by slow speeds for safety, but offers an impressive experience with open road testing and sensor fusion technology.
- The speaker discusses the use of a long reach excavator for protecting expensive equipment and mapping a golf course with GPS technology.
- The speaker discusses the limitations and capabilities of Tesla's Full Self-Driving system, highlighting the use of sonar technology and the need for slow speeds to ensure safety.
- The speaker discusses their experience driving a Tesla FSD for the first time and compares it to their experience with older vehicles.
- The speaker discusses their initial discomfort and concerns about using Tesla's Full Self-Driving feature, expressing amazement at the open road testing and lack of traditional controls.
- The speaker discusses the process of sensor fusion and grid generation for autonomous driving, noting that while it worked well at lower speeds, there may be more errors at higher speeds.
- The speaker discusses the challenges of navigating cluttered and unstructured environments with autonomous vehicles, the cost of hardware development, and the use of military bases for urban warfare training.
-
35:11 🚗 Tesla FSD requires calibration, individuals may need personalized solutions, but ultimately, widespread adoption of self-driving technology could lead to a common solution and potentially eliminate traffic lights.
- The speaker has extensive experience with autonomous vehicles, including driving and programming, and has purchased Tesla's Full Self-Driving feature for their Model Y, which required calibration before being able to function.
- One person is cautious while driving, while the other is more aggressive and takes risks.
- A personalized full self-driving solution may be necessary for each individual, but once it becomes fully autonomous, people may not care about the driving style.
- Humans are training the AI in three phases, tuning it for different locations and cultures, with the goal of achieving widespread comfort and adoption of self-driving technology in the future.
- Train humans to adapt to a common self-driving solution, maximizing road use and potentially eliminating traffic lights at intersections.
-
40:29 🚗 Tesla's full self-driving technology has the potential to significantly improve road safety, but challenges in achieving autonomous driving still exist.
- Cars with full autonomous hardware may not be required on the road until 2040, but there could be restrictions on human drivers in certain areas for safety and efficiency.
- Cars have a long lifespan, and if every car had Tesla's full self-driving, the roads would be safer and autonomous driving would work better.
- Tesla's full self-driving technology has the potential to significantly reduce fatalities and improve safety on the road, especially when compared to human drivers.
- The speaker discusses the challenges of unmanned systems and the use of rental cars in the DARPA challenge to create dynamic obstacles for autonomous cars.
- The speaker discusses their experience testing Tesla's Full Self-Driving (FSD) capabilities and the challenges of achieving autonomous driving.
- The speaker has only driven the car three times on full self-drive and has been predicting autonomous driving to be 5-10 years away since the 90s.
-
46:45 🚗 Tesla's Full Self-Driving has improved but the speaker is still nervous about its capabilities, prioritizing safety and suggesting it will take 15-20 years to achieve completely autonomous driving on the roads.
- Tesla's Full Self-Driving has shown significant improvement since 2008, but the speaker still feels nervous about its capabilities and believes it will take more time to become fully comfortable with it.
- The speaker discusses their experience with Tesla's Full Self-Driving feature and how their temperament affects their comfort with the technology.
- Safety is the top priority when optimizing for autonomous driving, with considerations for trip length and passenger comfort, as driving behavior is modified based on the presence of passengers.
- The speaker discusses the different driving environments and the potential for full self-driving solutions, suggesting that it will take 15 to 20 years to achieve completely autonomous driving on the roads.
- The speaker discusses the potential of life extension technology and compares the development of full self-drive technology to the perpetual 20-year timeline of nuclear fusion.
- Tesla's Full Self-Driving (FSD) could potentially allow for autonomous driving in limited geographies and conditions, with the hope of eventually being able to drive on freeways without having to pay attention to the road.
-------------------------------------
Duration: 0:54:6
Publication Date: 2024-04-15T17:07:29Z
WatchUrl: https://www.youtube.com/watch?v=gZJ-8JSa04A
-------------------------------------