Personally, I have never liked driving a semi-automatic car.
Semi-automatic car, I believe, is a mode that is useful for incompetent hill drivers who crash back while driving up the car parked uphill; semi-automatic car mode is useless other than that.
I hate that, in semiautomatic cars, it creeps, and the human driver needs to stop car by pressing brake pedal (also the pedal to increase engine speed on some occasions). I hate using the same brake pedal both for brakes, and somehow oppositly for engine mode speeding.
If the car engineer is looking to learn from his self driver's experience and create a robotaxi, or a Full-Self-Driving car stop-mode, he/she must first know how to drive the very oldest type of manually geared car, then seek perfection in fsd driving. Planning a fsd software improvement from a semi-auto car software skeleton is stupid.
FSD is great as it is. But, which stop-mode is better when a sudden event requires to go down from 100 ml/ph FSD to 0 speed.
How should the engine of the car and the shaft merge to prevent somersault accidents.
Is there a need to slow the engine a little bit at a Nötr Mode then connect the engine to shaft brakes.
Since recent car's direct-drive engines send mechanical energy directly to parts to move, sudden stop of engine is sudden stop of entire car at P or N mode.
Whereas, old cars allowed splitting working-engine and shaft.
In electric cars, a chemical way could be employed to stop engine, such as battery cooling in rare emergency events. Or some comparable ways other than mechanical brakes. I am only a driver liking safety not a car engineer to look at details of car speed energy.
There should be from high-speed (120 miles per hour) to zero speed factory experiments to see car behavior such as losing direction control, car somersault.
On a high way, to stop from 100 ml/ph to zero(full stop ), what should do the right hand of a human-driver? Pressing P button, or pulling a stalk; or, what hand-brake piece is extra needed for safe FSD stopping without any somersault.
Probably a stalk piece functioning for a strong brake without somersault somewhere for right hand of the human-driver is a good idea to stop a car from fsd mode.
I believe robotaxi FSD concept is almost solved for the average speeds such as 60 miles per hour, but higher speed braking safety problem is a big challenge to solve.
In summary, low speed and high speed fsd car stopping issues should have two different algorithms and include different mechanical pieces.
Any semiotomatic car foot brake pedal gear shifting algorithm is not necessary at all to study for Full Self Driving flow chart algorithm software.
As a low-speed lover driver, i believe low FSD car speeds are perfect for safety. However, on highways that impose an obligation to go fast, safe braking becomes more difficult as the speed gets higher, especially when passing other lane cars.
Another important issue is to apply correct speed at correct time. Though average low speed always is safer always while passing other car to leave it behind is an exception. Not being able to reach enough higher speeds because of false stupid high limit entry to car settings will cause accidents.
There should be two different maximal entry allowance at car settings (I believe it is there in Tesla comparable car fsd softwares, i did not look settings options yet). One will limit the speed in general except during passing other cars, other maximal speed setting will be for the scenario of passing another car.
If the software algorithm scenario is for surpassing another car, for changing the lanes, then software program should arrange maximal speed setting prioritizing the maximal speed of other car to surpass, and should adopt own car settings speed according to the measured speed of other car. Measurement of other car speed is done with similar mathematical equation built in software that is used to measure car own actual speed.
I believe Full Self Driving cars already is doing this at first by measuring the other car speed then increasing its own max speed at least 15 miles faster then the car to be surpassed.
PARKING AUTOPILOT SOFTWARE in FSD:
- I think it is not difficult to achieve perfection of FSD car park with Autopilot Software with current technology.
-The software already maps the parking lot, all other cars, objects, columns as coordinates, extracts the distances to itself, calculates its width, length, extensions like the mirrors of other parked cars inch by inch, shows the imaginary geometric tunnels it interrogates for parking.
-There are many excellent Geometry/Mathematics software programs to employ for park space computing. (Holographic Motion Capture Data from a human-drived parking car can be obtained one frame at a time and can be applied to problems)
At first, old fashioned human driver parks car into experimental places; car computer records its own size, its own geometry and creates best imaginary simulated tunnels, holographic paths to suggest.
-Car software compares whether the available spaces are suitable for its own maneuvering tunnels to enter the parking lot.
-An important issue is the controls of other cars in the parking lot, whether any part of the other car suddenly opens, whether there is a child in it, etc.
-Human-driver initiated Autopilot Park Option could touch any free space on touchscreen map, human-driver could ask parking calculations for the touched space fits for parking. There is requirement that human-driver uses his/her hand to suggest a free area on the touchscreen parking lot map. It is human drawing an elliptic area on touchscreen parking lot.
Such option cancels the requirement for parking lot white lines.
-Since parking areas are low-speed areas for all cars, i believe, Autopilot Parking Software is not a big issue to solve, or to improve.
-Car Full-Self-Drive option is very good already if it were not other cars to compute. That other cars movements are in general unpredictable is challenging . (For example, other car sudden speed change, sudden door opening, etc).
Translated by Google:
Özetle, robotaxi FSD konseptinin saatte 60 mil gibi ortalama hızlar için neredeyse çözüldüğüne inanıyorum, ancak daha yüksek hızlarda frenleme güvenliği sorunu çözülmesi gereken büyük bir zorluk.
Düşük hız ve yüksek hız fsd araba durdurma sorunları iki farklı algoritmaya sahip olmalı ve farklı mekanik parçalar içermelidir.
Herhangi bir yarı otomatik araba ayak freni pedalı vites değiştirme algoritması, Tam Kendi Kendine Sürüş akış şeması algoritma yazılımını çalışmak için hiç gerekli değildir.
Düşük hızı seven bir sürücü olarak, düşük FSD araç hızlarının güvenlik için mükemmel olduğuna inanıyorum. Ancak, hızlı gitme zorunluluğu getiren otoyollarda, hız arttıkça, özellikle diğer şeritteki arabaları geçerken güvenli fren yapmak zorlaşır.
Bir diğer önemli konu da doğru zamanda doğru hızı uygulamaktır. Ortalama düşük hız her zaman daha güvenli olsa da diğer arabayı geçerken onu geride bırakmak bir istisnadır. Araba ayarlarına yanlış aptal yüksek limit girişi nedeniyle yeterince yüksek hızlara ulaşamamak kazalara neden olur.
Araba ayarlarında iki farklı maksimal giriş izni olmalı (Tesla karşılaştırılabilir araba fsd yazılımlarında var sanırım, ayar seçeneklerine henüz bakmadım). Biri diğer arabaları geçerken genel olarak hızı sınırlayacak, diğer maksimum hız ayarı başka bir arabayı geçme senaryosu için olacaktır.
Yazılım algoritması senaryosu başka bir arabayı geçmek, şerit değiştirmek içinse, yazılım programı maksimum hız ayarını diğer arabanın maksimum hızına öncelik vererek düzenlemeli ve kendi araba ayar hızını diğer arabanın ölçülen hızına göre almalıdır.
Diğer araba hızının ölçümü, arabanın kendi hızını ölçmek için kullanılan yazılımdaki benzer matematiksel denklemle olabilir.
Tam Kendi-Kendine-Sürüş arabalarının bunu zaten, ilk başta diğer arabanın hızını ölçerek, ardından kendi maksimum hızını, aşılacak arabadan en az 15 mil daha hızlı artırarak yaptığına inanıyorum.