User Stories

Sumitomo Heavy Industries Speeds Development of Embedded Model Predictive Control Software for Hydraulic Excavators


Accelerate the design and implementation of embedded engine control software for hydraulic excavators


Use Model-Based Design with Simulink and Embedded Coder to model, simulate, and generate code from a model predictive controller that maximizes excavator work performance


  • Fuel efficiency increased by 15%
  • Engineering effort reduced by 50%
  • Aggressive deadline met

“With Model-Based Design, our hydraulics engineers can complete the controller design and implementation without involving an embedded engineer. That’s a big advantage, because it saves time and results in a higher-quality controller.”

Eisuke Matsuzaki, Sumitomo Heavy Industries
A Sumitomo hydraulic excavator.

A Sumitomo hydraulic excavator.

Hydraulic excavators need controls that keep the engine at a constant speed as the excavator picks up and drops off heavy loads. When designing these embedded controllers, engineers must take sudden fluctuations in load into account, as well as ensure that the design meets tightening emissions standards and safety requirements.

Facing pressure to meet new emissions regulations, Sumitomo Heavy Industries engineers used Model-Based Design with Simulink® to accelerate the design and implementation of hydraulic excavator control software.

“With Model-Based Design, we focused on developing control algorithms, not writing low-level code,” says Eisuke Matsuzaki, engineer at Sumitomo Heavy Industries. “As a result, we were able to deliver a high-quality controller in a limited amount of time while under tremendous time pressure.”


Sumitomo Heavy Industries, like many heavy equipment manufacturers, sources its engines from another supplier. As a result, the engineering team needed a way to determine the characteristics of the engine and its black-box RPM controller. They then needed to design a model predictive controller (MPC) that minimized changes in engine speed during load fluctuations while managing the delays associated with CAN bus communication. The team chose MPC because it accounts for input and output constraints and because they could adapt it to multiple engine types by changing the internal prediction model.

In the past, Sumitomo Heavy Industries followed a traditional development approach in which a control algorithm was designed by a hydraulics engineer, coded by an embedded engineer, implemented on a target processor, and then tested by the hydraulics engineer on a real excavator. Miscommunication between the engineers, coding errors, and limited access to test hardware often slowed development iterations. With stricter emissions standards about to take effect, the Sumitomo team needed to shorten development iterations and deliver a new controller on a tight schedule.


Sumitomo Heavy Industries engineers used Model-Based Design with Simulink to develop an embedded MPC for hydraulic excavators.

Working in Simulink with Simulink Design Optimization™, they built a transfer function–based plant model by estimating parameters from engine test data. They incorporated input disturbances such as the torque load placed on the hydraulic pump, as well as non-linearities such as power source output limits and dead time.

They then used the estimated plant model as an internal prediction model in an MPC designed with Model Predictive Control Toolbox™. They tuned controller parameters and verified the controller design via simulations in Simulink.

Using Embedded Coder®, the team generated hundreds of lines of C code from the controller model for a 32-bit RISC microcontroller. They updated the generated code to support multiple engine types with a single controller.

After completing performance evaluation tests on an actual excavator, Sumitomo Heavy Industries began using the new controller in production.


  • Fuel efficiency increased by 15%. “Sumitomo Construction Machinery achieved a 15% reduction in fuel consumption without sacrificing the excavator’s dynamic performance,” says Matsuzaki. “The increase in efficiency was due, in part, to a 50% reduction in engine speed fluctuations made possible by Model Predictive Control Toolbox and our improved control design.”
  • Engineering effort reduced by 50%. “By enabling us to eliminate the inefficiencies of a traditional development approach and follow a consistent workflow from control design to deployment, Model-Based Design has unquestionably saved us time,” says Matsuzaki. “While it is difficult to measure exactly, we estimate that we have cut engineering hours in half with Model-Based Design.”
  • Aggressive deadline met. “If we hadn’t met our deadline, we would have been unable to comply with Tier-4 Final emissions standards in time to continue selling excavators,” notes Matsuzaki. “Modeling and simulating the controller in Simulink and then generating code with Embedded Coder enabled us to deliver on schedule because we kept our focus on the control algorithm rather than on writing code.”