When AP 2.0 hardware was released (along with a video showing a fully autonomous drive from a residential neighborhood to the Tesla Factory), Tesla made it sound like FSD was just around the corner. The Web page for FSD described it as if it was ready for release, even going so far as to warn customers that 'using a self-driving Tesla for car sharing and ride hailing for friends and family is fine, but doing so for revenue purposes will only be permissible on the Tesla Network'.
We all know now that Tesla has a long way to go to achieve FSD/Level 5. However, the recent autopilot enhancements in 9.0 (that were subsequently pulled) have given us an idea of how they might gradually introduce FSD features over time. Here are my (completely uninformed) predictions as to how the next few years might look:
1) Tesla releases Navigate on Autopilot, including automatic lane changes to get around slower vehicles or reach highway exits. This could be the last major EAP feature released. This is also (arguably) *very close* to full self driving, just limited to highways. While I know it may not handle every edge case, it will make it possible for drivers to hand off to Autopilot completely for long road trips, taking back control only after exiting the highway.
2) Tesla releases stop sign and red light detection when Autopilot is engaged. Just as Navigate on Autopilot brings us much closer to full autonomy on the highway, red light/stop light detection would enable drivers to leverage Autopilot for longer periods of time when not on the highway. This may be the first 'FSD' feature released.
3) At some point along the way, Tesla will start reducing the nag warnings, eventually eliminating them altogether. I think the nag warnings are only a stop gap (of limited actual value) that will only remain in place until everyone involved - Tesla, owners, and the general public - feel more confident in the capabilities of Autopilot.
At this point most of the daily driving tasks will be handled by the vehicle. The driver will be needed to get the car onto major roads, take turns at intersections and be available at a moment's notice to handle edge cases. Some might argue that this reliance on the driver is a major issue, but I still believe a system at this level has tremendous value. Even with EAP as it is today I find driving much more enjoyable and less stressful.
4) The next major FSD release will extend 'Navigate on Autopilot' to non-highway roads, including turning at intersections, making lane changes as needed, etc. This will not be a full L5 system that can handle every possible scenario without the need for a human driver. It won't, as detractors like to point out, be very good at handling winding dirt roads under construction with livestock wandering along the median. And it will still require the driver to be available to take over at a moment's notice. In particular, when it reaches its destination it will hand control back to the driver to park the vehicle. Finding a suitable place to park has always seemed like one of the most far-fetched FSD use cases - does it park in your driveway? On the street? How does it know which spaces are reserved for which restaurant? How does it get into a parking garage? etc.
5) Now we're left with closing that last 10% to achieve full L5 and eliminate the human driver. This part is honestly very hard to imagine/predict. And I certainly have zero expertise in this field (unlike many of the contributors to this forum). I have to believe that it's possible (or else everyone in this field is delusional), but I think it may take many many years to get there.
Note that steps 1-4 all require a human driver, and (gasp) a steering wheel. Modern aircraft handle much of the drudgery of commercial aviation, but there are still controls for the pilot and co-pilot, who must be available to take over at a moment's notice.
Even if step 5 never happens (or takes a decade), I guess the point I'd like to make is that there is still an incremental path towards FSD that will make our cars safer and that much more fun to 'drive', even if we're not the ones doing most of the driving
We all know now that Tesla has a long way to go to achieve FSD/Level 5. However, the recent autopilot enhancements in 9.0 (that were subsequently pulled) have given us an idea of how they might gradually introduce FSD features over time. Here are my (completely uninformed) predictions as to how the next few years might look:
1) Tesla releases Navigate on Autopilot, including automatic lane changes to get around slower vehicles or reach highway exits. This could be the last major EAP feature released. This is also (arguably) *very close* to full self driving, just limited to highways. While I know it may not handle every edge case, it will make it possible for drivers to hand off to Autopilot completely for long road trips, taking back control only after exiting the highway.
2) Tesla releases stop sign and red light detection when Autopilot is engaged. Just as Navigate on Autopilot brings us much closer to full autonomy on the highway, red light/stop light detection would enable drivers to leverage Autopilot for longer periods of time when not on the highway. This may be the first 'FSD' feature released.
3) At some point along the way, Tesla will start reducing the nag warnings, eventually eliminating them altogether. I think the nag warnings are only a stop gap (of limited actual value) that will only remain in place until everyone involved - Tesla, owners, and the general public - feel more confident in the capabilities of Autopilot.
At this point most of the daily driving tasks will be handled by the vehicle. The driver will be needed to get the car onto major roads, take turns at intersections and be available at a moment's notice to handle edge cases. Some might argue that this reliance on the driver is a major issue, but I still believe a system at this level has tremendous value. Even with EAP as it is today I find driving much more enjoyable and less stressful.
4) The next major FSD release will extend 'Navigate on Autopilot' to non-highway roads, including turning at intersections, making lane changes as needed, etc. This will not be a full L5 system that can handle every possible scenario without the need for a human driver. It won't, as detractors like to point out, be very good at handling winding dirt roads under construction with livestock wandering along the median. And it will still require the driver to be available to take over at a moment's notice. In particular, when it reaches its destination it will hand control back to the driver to park the vehicle. Finding a suitable place to park has always seemed like one of the most far-fetched FSD use cases - does it park in your driveway? On the street? How does it know which spaces are reserved for which restaurant? How does it get into a parking garage? etc.
5) Now we're left with closing that last 10% to achieve full L5 and eliminate the human driver. This part is honestly very hard to imagine/predict. And I certainly have zero expertise in this field (unlike many of the contributors to this forum). I have to believe that it's possible (or else everyone in this field is delusional), but I think it may take many many years to get there.
Note that steps 1-4 all require a human driver, and (gasp) a steering wheel. Modern aircraft handle much of the drudgery of commercial aviation, but there are still controls for the pilot and co-pilot, who must be available to take over at a moment's notice.
Even if step 5 never happens (or takes a decade), I guess the point I'd like to make is that there is still an incremental path towards FSD that will make our cars safer and that much more fun to 'drive', even if we're not the ones doing most of the driving