Could you tell a Tesla what you are thinking about its actions through a MindWave Mobile 2 EEG headset? I think it could help train the Tesla what would do when you see a stop sign or red traffic light. I am pretty good at anticipating when a green light is about to change to yellow, I'm sure that my brain waves could be communicated to the Tesla and would intern have the Tesla adjust its speed.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
FSD - Could a Tesla communicate with a human with a headset that picks up human wave links?
- Thread starter Knight205
- Start date
Silicon Desert
Well-Known Member
smogne41
Active Member
My Tesla already adjusts it's speed based on what I am thinking. Because my brain controls the pedals (using me feet).
Implant a neural link chip?
Neuralink
Creating a generalized brain interface to restore autonomy to those with unmet medical needs today and unlock human potential tomorrow.
neuralink.com
Good god, can you imagine what Elon is thinking about when he's in a car? Me neither, but I'm sure it doesn't translate to good safe driving. For that matter probably very few of us have control over our thoughts enough to control the car's actions. Obviously we have some kind of brain to body filter that mostly prevents us from acting on our impulses - except giving the finger and stuff. I know I still have this stupid "pedestrian running across the road - worth 10 points" thing that goes through my head from grade school or something, I'm not sure why it's there, but I don't act on it, some part of the brain filters that out. I'm not sure if a mind-reading headset would do well. That would be a pretty lame defense in court "Sorry I didn't mean to run the red light, my brain told the car to do it".
Lol. Ok, assuming you actually statistically are able to do this better than guessing, I'm still not quite sure I see the point.
Do you drive differently approaching a stale green? If so... faster? Or slower?
The car should be able to make instantaneous and mathematically perfect go / no-go decisions when light turns yellow (perfect meaning stop, if possible, without exceeding acceleration, jerk, and time-to-clear intersection constraints). I don't see the point in trying to predict it in advance. There's such minimal room for that lead to better driving decisions and plenty of opportunity for any incorrect prediction to cause unnatural behaviors.
I have a different proposal: Tesla needs to change the UI to display what the car "thinks" and will do before it takes the real action.
Example: when the car creeps to the creeping line to make left or right turn, the software should display an icon/message to indicate the car decides to take the actual turn. This will make me feel safer and give me a chance to disengage if the the car has an unsafe decision. There are many other things the UI should tell us what the car is thinking.
Example: when the car creeps to the creeping line to make left or right turn, the software should display an icon/message to indicate the car decides to take the actual turn. This will make me feel safer and give me a chance to disengage if the the car has an unsafe decision. There are many other things the UI should tell us what the car is thinking.
Yes.
Also we need the software tell us whether the car does what it's supposed to do. Right now, FSD displays "stopping for stop sign". It displays that message after it begins to slow down. It's too late. We want to know whether the car recognizes the stop sign or not first. If the car moves close to the stop sign but we don't know whether the car sees it or not then we disengage. And pedestrians, potential accident causing objects,...
Last edited:
Okay, maybe the spirit of my thoughts was lost! The simple idea was to share what appropriate behavior is in your thoughts with the computer's actions to modify its behavior to anticipate how to respond to the light's potential state in the future like is the person going to step off the curb and be in your path, do you slow down or make an evasive maneuver. If you are going slower below the set speed limit due you increase speed or begin slowing down. it is not illegal the pass under a yellow light.
When/if FSD Beta goes L3 or higher you can NOT control the car in any way since you are NOT driving. The most you can do is make a request of the car but the car "decides" all driving behaviors. If you take control of the car in any way the system MUST disengage to at least L2.
I don't think there is a mechanism to downgrade an SAE level. If the system is on L3 and you disengage the system (using the brake pedal for example), it returns to L0, with the driver in total control. You can then initiate L2 if you'd like, but the system won't walk down the tree.
That's amazing you can train your Tesla with your feet, but what's in your mind is not the same information as your feet, which is communication I'm thinking about. The robot "Optimus" must also be trained, but I guess your feet would train that too.
I'm not suggesting control, only your thought input to help the computer be more human-like with greater safety than a human - autopilot still slows down for orange barrels on the side of the road and is slow to accelerate when the light turns green and the car in front has moved away much human- like.
Yes, Elon: "How can I make this car work for humans, and what is the quickest way to accomplish that task?" The computer has to work in a non-perfect human world.
I'm aware of "Neural Link Chip" but my concept is the thoughts you have would simply be additional input for the computer to learn how to deal with humans. I'm not trying to have a two-way conversation or control, so no chip is needed.
I'm not trying or even suggesting controlling the Tesla why everyone thinks train means control. The training computer Dojo should only use data that is of value to train the computer, what the aggregate human thought norm is, not a specific speed up, slow down, watch out, that car is about to cut you off.
Similar threads
