Algorithm to detect and mitigate real-time perceptual adversial attacks on autonomous vehicles

A tangible, non-transitory, machine-readable medium, comprising machine-readable instructions, that when executed by a processor, cause the processor to:

• receive input data, wherein the input data includes sensor data from a plurality of sensors and map data;
• monitor, in real time, an environment around an autonomous vehicle to identify a region of the environment around the autonomous vehicle that is possibly subject to an adversarial attack;
• determine a probability of the adversarial attack in the region of the environment around the autonomous vehicle that is possibly subject to the adversarial attack;
• determine whether the probability of the adversarial attack in the region of the environment around the autonomous vehicle that is possibly subject to the adversarial attack is greater than a predetermined threshold;
• in response to determining that the probability of the adversarial attack in the region of the environment around the autonomous vehicle that is possibly subject to the adversarial attack is greater than the predetermined threshold, plan a motion of the autonomous vehicle by taking into account the adversarial attack to generate a planned motion;
• and
• control the autonomous vehicle to move in accordance with the planned motion.

A system for an autonomous vehicle, comprising:

• a plurality of sensors;
• a controller in communication with the plurality of sensors, wherein the controller is programmed to:
• receive input data, wherein the input data includes sensor data from a plurality of sensors and map data;
• monitor, in real time, an environment around the autonomous vehicle to identify a region of the environment around the autonomous vehicle that is possibly subject to the adversarial attack;
• determine a probability of the adversarial attack in the region of the environment around the autonomous vehicle that is possibly subject to the adversarial attack;
• determine whether the probability of the adversarial attack in the region of the environment around the autonomous vehicle that is possibly subject to the adversarial attack is greater than a predetermined threshold;
• in response to determining that the probability of the adversarial attack in the region of the environment around the autonomous vehicle that is possibly subject to the adversarial attack is greater than the predetermined threshold, plan a motion of the autonomous vehicle by taking into account the adversarial attack to generate a planned motion;
• and
• control the autonomous vehicle to move in accordance with the planned motion.