A robot can not cause harm or by inaction allow harm to humans?

A robot can not cause harm or by inaction allow harm to humans? And if the robot is in a hopeless situation.?

For example, a car drives on the highway on autopilot, and suddenly a person jumps from the sidelines. Given the rather high speed of the car, let’s assume that any maneuver in this section of the road will lead to the death of the driver or pedestrian. What should an autopilot do?

Hypothetically - it will go into a logical cliff and die trying to solve an insoluble ethical dilemma (see, for example, St. Lem's short story “Liar” from the classic cycle “I am a Robot”).

At present, an AI of the level capable of being guided by ethical considerations set forth in the four laws of robotics has not yet been created. One example of the lag of achievements of current technology from science fiction

If the robot is in a hopeless position, then you must judge the designer of the robot, as well as those who took it into operation.

They did not fulfill their duties for which they are paid money.

And this is regardless of whether the person was jumping or not jumping.

You mean THREE LAWS OF ROBOTICS. Any law restricts and restricts certain meanings. In the situation you are describing, the designers of the moving car are to blame and the one who accidentally or intentionally left / got / got in the way of the vehicle. I consider the optimal answer to the question: to ensure maximum safety for both the driver and the pedestrian by separating the path of the pedestrian and the vehicle.

I have bad news for you - maybe. Some of them have been created for this, while others don’t give a damn about someone who, for example, is under his welding machine.

There is such a feature in an Tesla electro car. The pilot himself chooses in the settings how to drive a car. Knock down a person or die.