Wireless Actuator Control Using Fallback Commands for Packet Loss
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Wireless communication systems in control applications are prone to packet losses, which can lead to instability and performance degradation in dynamical systems controlled by PID or MPC controllers, especially when nominal commands are not received, potentially causing critical situations in applications like autonomous vehicles or robots.
Innovation Solution
A method is proposed where a controller sends both a nominal command and a switching command, with the switching command being used in case of communication blackouts or data corruption, allowing the actuator to maintain operation and drive the system to a predicted state using resilient MPC algorithms and chance constraints, ensuring safety and performance even during rare events.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If wireless communication is used for control applications, then mobility and flexibility are improved, but packet losses occur due to radio propagation effects
Solution Approach 1:
The controller pre-calculates and transmits multiple future control commands to the agent before communication blackouts occur. This preliminary action ensures that the agent has a buffer of commands to execute during packet losses, maintaining system reliability while preserving wireless mobility.
Solution Approach 2:
The system transmits redundant control commands in advance to cushion against the harmful effect of packet losses. By having multiple pre-sent commands, the agent can continue operation during communication failures, protecting against the unreliability of wireless transmission.
2Ease of manufacture
If PID controller is used, then ease of implementation is improved, but optimal control is not achieved and hunting occurs
Solution Approach 1:
The system transitions from fixed PID parameters to dynamic MPC parameters that adapt based on predicted future states and communication conditions. This parameter change enables optimal control while maintaining implementation feasibility through the use of pre-calculated commands.
Solution Approach 2:
The MPC controller uses feedback from current system state and communication status to dynamically adjust control commands. This feedback mechanism eliminates the hunting phenomenon by continuously adapting control actions to actual system conditions rather than relying on fixed proportional-integral-derivative relationships.
3Reliability
If MPC controller is used, then optimal control is achieved, but computational complexity increases
Solution Approach 1:
The controller performs computationally intensive MPC calculations in advance to generate a sequence of future control commands. By pre-computing these commands when communication is reliable, the system reduces real-time computational burden while maintaining optimal control performance through the use of model predictive algorithms.
4Productivity
If communication blackouts occur, then system safety is compromised, but continued operation is needed
Solution Approach 1:
The system pre-calculates and transmits control commands that account for potential communication blackouts and safety constraints. These pre-planned commands ensure the agent can maintain safe operation during packet losses without compromising productivity or requiring real-time computation during critical periods.
Solution Approach 2:
The controller pre-computes compensatory commands that counteract the potential harmful effects of communication failures. By anticipating packet losses and preparing corrective actions in advance, the system maintains safety constraints while ensuring continued operation during blackouts.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
The description proposes a method to control an agent (AGT) operating at least one actuator according to commands repeatedly received from a controller (CTLR), said agent receiving repeatedly measurement data from at least one sensor sensing an environment of the actuator, said measurement data defining, at an iteration k, a current state xkn of the actuator's environment resulting from the application by the actuator of a nominal command uk−1n at a previous iteration k-1, said nominal command uk−1n at previous iteration k-1 being initially: - computed by the controller (CTLR) based on a current state xk−1n of the actuator's environment defined at least by measurement data acquired by the sensor during said previous iteration k-1, and - transmitted by the controller (CTLR) to the agent (AGT) to be applied by the actuator, Wherein the controller (CTLR), at each iteration k: - computes, in addition to said nominal command ukn, a switching command uk+1c based at least on the current state xkn at said iteration k, and - transmits to the agent (AGT) said switching command uk+1c along with said nominal command ukn, said switching command uk+1c being intended to be used by the agent to operate the actuator in case no command is received by the agent following the application of the nominal command ukn of iteration k.