Dwindra Sulistyoutomo

Portfolio

Selected Projects

Solution for manual labor dependency

MHI Automated Picking Solution

Mitsubishi Heavy Industries, Ltd. (MHI) has developed an automated picking system that integrates Automated Guided Forklifts (AGFs), Automated Guided Vehicles (AGVs) and palletizer robots. This solution addresses the dependency on manual labor and the challenges of lifting heavy items.

As the robotics engineer from Hitachi Automation, Ltd., I contributed to this project by developing the KRC Palletizer Robot system and overseeing the integration with the Warehouse Control System (WCS).

The system has been successfully deployed and is now operational at a live logistics site in partnership with Kirin Group Logistics Co., Ltd., demonstrating its effectiveness in real-world conditions.

External links:

MHI and Mitsubishi Logisnext to Commence Provision of Automated Picking Solutions for Warehouses Based on "ΣSynX"
https://www.mhi.com/news/22083102.html
Joint Demonstration of Automated Picking Solution Launched in Collaboration with Kirin Group
https://www.mhi.com/news/22112101.html
Full Implementation at Kirin Group's Ebina Logistics Center, the First Such System in Japan
English: https://www.mhi.com/news/241212.html
Japanese: https://www.mhi.com/jp/news/241212.html

Palletizing Robot

Intelligent automation solutions for logistics

KRC Palletizing Robot

Hitachi Automation, Ltd. delivers cutting-edge automation solutions powered by intelligent robotics. Under the Kyoto Robotics (KRC) brand, our mission is to create a society where robots manage labor-intensive tasks, reducing the burden of heavy work through innovative palletizing and depalletizing systems.

For over five years, I have been working on developing KRC palletizing robot system.
Currently, I act as the primary technical expert, overseeing and directing software development for these systems. I am mainly in charge of the motion planning and control algorithms for robot arms.

In addition to robot system development, I designed a tree-based stacking algorithm for box palletizing. This algorithm enables efficient palletizing, supporting SKU-based order palletizing, random-box order palletizing, and unordered palletizing.

I also contributed as the lead designer for the web-based GUI for the ROS-based system and developed a CI/CD pipeline to improve the maintenance of ROS-based repositories.

This system has been patented under Hitachi Automation, Ltd.
Patent: Workpiece Stacking System (Patent No: 特許第 7387931 号 (P7387931))

External links:

Patent: Workpiece Stacking System (Patent No: 特許第 7387931 号 (P7387931))
Kyoto Robotics Brand website
English: https://www.hitachi-automation.co.jp/kyotorobotics/en/
Japanese: https://www.hitachi-automation.co.jp/kyotorobotics/jp/
KRC Palletizing Robot (Japanese only)
https://www.hitachi-automation.co.jp/kyotorobotics/jp/palletize-solution
Kyoto Robotics Brand Promotional video https://www.youtube.com/watch?v=g80G7OZZISQ

Mini SLAM Robot

Small, powerful, low-cost SLAM robot with ROS support

Mini SLAM Robot

SLAM (simultaneous localization and mapping) is essential for autonomous robots. With the availability of low-cost sensors, it is possible to build our own SLAM inexpensively. The main objective of this project is to build a reliable mobile robot platform for other mobile robot projects, like reinforcement learning.

In this project, I built a mini LiDAR-based SLAM robot. The robot uses a 4WD Robot Kit that includes robot chassis, 4 DC motors, DC motor driver, Raspberry Pi Camera, and ultrasonic sensor. The robot kit has a customization feature, and it is possible to mount external LiDAR on top of the robot body. However, the kit has a limitation that only Raspberry Pi boards are supported. For the LiDAR, I use YDLIDAR X2L for its small size yet good range—perfect for indoor robots and development—, and it is a perfect fit on top of the robot. Furthermore, the robot has MPU6050 IMU mounted for an additional sensor.

The chosen lidar sensor supports ROS-based development (see YDLIDAR ROS package). To process the data, I use the Hector-SLAM ROS package. Hector-SLAM is an algorithm that can be used without odometry and only requires lidar scan data.

Servo Robot Arm

ROS driver for 6DOF Servo Robot Arm based on ROS-Industrial

Servo Robot Arm

Based on the library created by ROS-Industrial, it is possible to develop any ROS driver for an industrial robot arm.
In this project, I created a ROS driver for a servo robot. This kind of robot is commonly known—one can buy it easily anywhere—that uses servo motors as its actuators. The purpose is to make it possible for anyone wishing to try the ROS-Industrial library, even on a simple servo robot.

Currently, the ROS driver works on a 6DOF Servo Robot Arm. The MoveIt configuration is also already set according to the 3D model defined on the robot description.
See repository for ROS Driver

On the robot side, a ROS controller is designed to handle the communication with the ROS Driver. The ROS Controller includes Simple Message as the main communication protocol from ROS-Industrial. The ROS Controller manages joint commands and robot status messages based on that protocol.
See repository for ROS Controller on robot side

Musculoskeletal Robot

A bio-inspired humanoid robot using artificial muscles with agile locomotion skill

Agile Musculoskeletal Robot

We introduced a novel approach to realizing dynamic motion on a humanoid robot. A combination of various dynamic motions may lead to better performance of dynamic motion tasks by utilizing the energy transfer between the motions. We have developed a robot capable of doing sequential jumping-stepping motion for agile locomotion.

A bio-inspired robot was developed based on the human musculoskeletal system in order to realize human dynamic motion on a robot. Pneumatic artificial muscle was chosen as the actuator to realize explosive movements.

For realizing the motion, we decided muscle control command for each motion referring to the muscle activation pattern of a human. We have conducted experiments for the sequential jumping-stepping motion with our motion strategy on the developed musculoskeletal robot. The results show that the musculoskeletal robot performs faster stepping by performing pre-jumping before a stepping than stepping without pre-jumping.

The article was published in Proceedings of IEEE International Conference on Robotics and Biomimetics (ROBIO 2018).
See article

Any notes related to this work were documented in Blog posts.

Musculoskeletal Force Analysis

Force analysis tool for a musculoskeletal system

Musculoskeletal Force Analysis

Based on the Maximum Output Force Profile (MOFP) [1], we developed a force analysis tool for a musculoskeletal system. The force analysis can be used as a framework to design a musculoskeletal robot with pneumatic artificial muscles. The tool includes GUI to analyze the dynamics interactively.

The tool currently only supports analysis for a musculoskeletal leg. Nevertheless, the force analysis for a musculoskeletal arm can be computed based on this tool. Please check the reference for the details.

The force analysis provides calculation for both muscles with constant moment arms and angle-dependent moment arms [2]. Experiment settings and data from the cited papers are provided for references.

The force analysis tool is available at the Github repository.

For further description of the analysis, please check the notes on the Force Profile post.

References:
[1] R. Niiyama and Y. Kuniyoshi, “Design principle based on maximum output force profile for a musculoskeletal robot”. Industrial Robot: An International Journal, 2010.
[2] S. Nishikawa, K. Tanaka, K. Shida, T. Fukushima, R. Niiyama, Y. Kuniyoshi, “A musculoskeletal bipedal robot designed with angle-dependent moment arm for dynamic motion from multiple states”. Advanced Robotics, vol. 28, no. 7, pp. 487-496, 2014.

RoboSoccer

Faster response and improved performance using Anti-Windup Speed Control System

Anti-Windup Speed Control System for RoboSoccer

RoboCup is a robotics competition that enhances robotics research in an appealing challenge: a soccer game. RoboCup Small-size league is one of the leagues which focuses on a multi-agent system in a highly dynamic environment.

Each robot agent has a low-level controller to handle every command from the server PC. As the environment is highly dynamic, an excellent agent needs a proper control system. The main challenge for the controller is to have a fast response to speed and direction commands with high frequencies. Hardware limitation becomes another challenge in designing the proper controller.

A PI anti-windup control system was chosen to increase the system response and handle integral windup caused by actuator limitation. The designed controller improved the rise time up to five times, making the agent agile and suitable for the dynamic environment.

The article was covered in Tokyo Tech Indonesian Commitment Award 2014.
See article

RFID-based Driving License Card

A conceptual design for traffic jam solution

RFID-based Driving License Card

RFID-based Driving License Card is a conceptual design for a traffic jam solution.

Jakarta is one of the most congested cities, with its main problem being the overutilization of private vehicles. The government establishes the odd-even license plate restriction system to solve the problem. However, rule violation keeps taking place where a lot of cars that are not supposed to be on the road keep running on the road.

We proposed an idea to increase the system's effectiveness by utilizing a digital license plate. The license plate shows the car's status, whether it is in the scheduled zone on the day (based on the odd-even system).

The system also involves the driver's license card for additional features. We proposed an idea to implement RFID on the driver's license card. By using that, the license plate also shows a driver who gets a traffic ticket. So, the traffic officers may know a repeated violation quickly. Also, it makes it possible for a case where the system allows a driver to ask for special permission to drive beside the designated day (e.g., odd number car on even date).

The project was done in collaboration between Institut Teknologi Bandung and Kyungpook National University for IS-CAPS (International Students Joint Capstone Design).

Arduino Yun + Android

Interfacing Arduino Yun using an Android program over WiFi

Arduino Yun + Android

Arduino Yun was the first Arduino board with a WiFi module. As this was a new great feature for an Arduino at that time, it was interesting to try it out. The project was simple interfacing the Arduino GPIO over WiFi.

In addition, it was also interesting to make the interface from a mobile app to add more IoT feels. So, I also created an Android app as the interface to control the Arduino GPIO.

Dwindra Sulistyoutomo

Robotics Engineer & Researcher

About

Research Interests and Background

Publications

Conference Papers

Patents

Portfolio

Selected Projects

Archived Projects

Blog

Skills

Design

Platforms

Programming languages