Systems and methods for autonomous vehicle behavior control

A vehicle control system, comprising:

• one or more sources of input data, comprising at least one of:
• sensors that provide sensor output information, map data and goals;
• a low-level controller configured to process control signals to generate commands that control a plurality of vehicle actuators of a vehicle in accordance with the control signals to execute one or more scheduled actions to be performed to automate driving tasks;
• and
• a high-level controller comprising a plurality of first controller modules, and being configured to generate the control signals, wherein the first controller modules of the high-level controller comprise:
• an input processing module configured to generate symbolic and sub-symbolic percepts based on the input data;
• a memories module configured to generate, based on the symbolic and sub-symbolic percepts, memories data, state operations and first feedback information that is provided to each of the other first controller modules;
• a world model module configured to generate, based on the symbolic and sub-symbolic percepts, world state information that indicates state of other entities in an external environment of the vehicle and includes information regarding current attributes of objects in the external environment of the vehicle, and second feedback information that is provided to each of the other first controller modules;
• a decision processing module configured to generate, based on the symbolic and sub-symbolic percepts, queries, state operations, memory operations, a control model comprising control actions and constraints, and third feedback information that is provided to each of the other first controller modules;
• and
• a projection module configured to:
• process projection inputs to generate:
• memory operations that are sent to the memories module to update the memories module;
• state operations that are sent to the world model module to make revisions at the world model module;
• projections that are used to make a behavior decision at the decision processing module;
• and fourth feedback information that is provided to each of the other first controller modules, and wherein the decision processing module comprises:
• a control model executor module that is configured to:
• generate, by executing the control model, the control signals that are provided to the low-level controller, wherein the control signals specify control actions and parameters used to schedule the one or more scheduled actions to be performed to automate driving tasks.

A high-level controller configured to generate control signals that are used to generate commands that control a plurality of vehicle actuators of a vehicle to execute one or more scheduled actions to be performed to automate driving tasks, the high-level controller comprising:

• a plurality of first controller modules, comprising:
• an input processing module configured to generate symbolic and sub-symbolic percepts based on one or more sources of input data, comprising at least one of:
• sensors that provide sensor output information, map data and goals;
• a memories module configured to generate, based on the symbolic and sub-symbolic percepts, memories data, state operations and first feedback information that is provided to each of the other first controller modules;
• a world model module configured to generate, based on the symbolic and sub-symbolic percepts, world state information that indicates state of other entities in an external environment of the vehicle and includes information regarding current attributes of objects in the external environment of the vehicle, and second feedback information that is provided to each of the other first controller modules;
• and
• a decision processing module configured to generate, based on the symbolic and sub-symbolic percepts, queries, state operations, memory operations, a control model comprising control actions and constraints, and third feedback information that is provided to each of the other first controller modules;
• a projection module configured to:
• process projection inputs to generate:
• memory operations that are sent to the memories module to update the memories module;
• state operations that are sent to the world model module to make revisions at the world model module;
• projections that are used to make a behavior decision at the decision processing module;
• and fourth feedback information that is provided to each of the other first controller modules, and wherein the decision processing module comprises:
• a control model executor module that is configured to:
• generate, by executing the control model, the control signals, wherein the control signals specify control actions and parameters used to schedule the one or more scheduled actions to be performed to automate driving tasks.