WB-toolbox OSQP block failing to converge with robotology-superbuild version >= v2025.05.0

[Gio-DS]

• I already updated to the latest version I can use
• I already checked similar issues using the search box
• I already checked the website for known troubleshooting
• I already cleaned my environment (by removing the wb-toolbox install folder and installing it again)
• I already excluded the possibility that system configuration files (e.g. yarp related) might be responsible
• I am sure the environment variables are correct
• I tried to reproduce the error in simulation (if it affects the real robot)

Description

As per title, the Simulink OSQP block does not work if used with robotology-superbuild with version v2025.05.0 or later. More specifically, the block outputs an incorrect solution (with solver status = 7) to a given set of optimization weight matrices and constraints, while the same optimization problem is correctly solved with robotology-superbuild v2025.02.0 (proving that the optimization problem is feasible).
No errors are displayed neither on the simulink diagnostic viewer nor in the terminal from which Matlab has been launched.
It seems to me that the problem comes from the inequality constraints: higher (in terms of absolute value) lower and upper bounds make the solver converge (solver status = 1) as the inequality constraints remain probably inactive.

I have also tried to change just the WB-toolbox version to v6.0.1 and v6.0.2 keeping the robotology-version at 2025.02.0, and the solver converges to a valid solution.

Platform

The same behavior is observed both in simulation and on the real robot that uses the generated code from that Simulink architecture.

Screenshots

Converging to solution, with nominal upper and lower bounds (robotology-superbuild v2025.02.0):

Not converging to solution, with nominal upper and lower bounds (robotology-superbuild v2025.05.0):

Converging to solution, with higher (in terms of absolute value) lower and upper bounds (robotology-superbuild v2025.05.0):

Additional information

• Ubuntu 22.04
• Matlab Version: 2025b

cc @valegagge

[traversaro]

Probably the regression is due to the update from osqp v0.6 (https://github.com/robotology/robotology-superbuild/blob/v2025.11.1/releases/2025.02.0.yaml#L9) to osqp v1 (https://github.com/robotology/robotology-superbuild/blob/v2025.11.1/releases/2025.05.0.yaml#L9).