Automatic Driving Robot Control Device Weighted Operation Sequences
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automatic driving robot control methods using learning models often result in abrupt operations due to sudden changes in inferred pedal operations, leading to decreased conformity to command vehicle speed and potential worsening of fuel economy and exhaust gas performance.
Innovation Solution
A control device and method that utilizes a running state acquisition unit, an operation content inference unit, and a vehicle operation control unit to calculate a weighted sum of past operation sequences, generating a control signal for smooth vehicle operation and high accuracy in following command vehicle speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If a learning model is used to infer pedal operations, then the automatic driving robot can operate the vehicle, but abrupt operations occur due to sudden changes in inferred operations
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing multiple candidate operation sequences before actual vehicle operation. The learning model generates several possible operation sequences in advance, and the system selects the most appropriate sequence based on current vehicle state, avoiding abrupt changes during real-time operation.
Solution Approach 2:
The patent implements dynamics by making the operation sequence selection adaptive to current vehicle conditions. The system dynamically chooses from multiple pre-calculated operation sequences based on real-time vehicle state, allowing smooth transitions between different operation patterns while maintaining responsiveness to changing conditions.
2Stability of the object's composition
If learning model inference results are amended by moving average or low pass filter, then operation smoothness is improved, but conformity to command vehicle speed deteriorates due to delay
Solution Approach 1:
The patent pre-calculates multiple operation sequences with different characteristics (including both smooth and responsive sequences) before operation. By having multiple pre-prepared sequences to choose from, the system can select the most appropriate one without needing to apply smoothing filters that would cause delays.
Solution Approach 2:
The system dynamically selects from multiple pre-calculated operation sequences based on current vehicle state and requirements. This allows the system to maintain both smoothness and speed conformity by choosing the most suitable pre-calculated sequence rather than applying smoothing that would cause delays.
3Speed
If abrupt pedal operations are applied to the drive robot, then the vehicle can respond quickly to speed commands, but fuel economy and exhaust gas performance worsen
Solution Approach 1:
The patent pre-calculates multiple operation sequences with different characteristics, including sequences optimized for fuel efficiency. By having these sequences prepared in advance, the system can select fuel-efficient operation patterns without sacrificing response speed.
Solution Approach 2:
The system dynamically selects operation sequences based on current vehicle state, allowing it to choose fuel-efficient sequences when appropriate while maintaining quick response when needed. This adaptive selection balances response speed and fuel economy based on real-time conditions.
Data Source
AI summary
[Problem] To provide an automatic driving robot control device and control method that enable a vehicle to be operated smoothly while also being caused to conform to a command vehicle speed with high accuracy.[Solution] The present invention provides an automatic driving robot (drive robot) 4 control device 10 that controls the automatic driving robot 4, which is installed in a vehicle 2 and causes the vehicle 2 to run, such that the vehicle 2 runs in accordance with a defined command vehicle speed, wherein the automatic driving robot 4 control device 10 is provided with: a running state acquisition unit 22 that acquires a running state of the vehicle 2 including a vehicle speed and the command vehicle speed; an operation content inference unit 31 that infers, on the basis of the running state, an operation sequence, which is a sequence of operations of the vehicle 2 at a plurality of times in the future that causes the vehicle 2 to run in accordance with the command vehicle speed, by using an operation inference learning model 40 that was trained by machine learning to infer the operation sequence; and a vehicle operation control unit 23 that extracts, from each of the operation sequences inferred a plurality of times in the past, the operations corresponding to a control time for subsequently controlling the automatic driving robot 4, calculates a weighted sum of these extracted plurality of operations to calculate a final operation value, generates, on the basis of the final operation value, a control signal for controlling the automatic driving robot 4, and transmits the control signal to the automatic driving robot 4.