Annals of a mechatronic system design project
By Professor Kevin C. Craig and Matthew A. Rosmarin
Rensselaer Polytechnic Institute
Figure 1. The Engineering System Investigation Process.
A senior design team at Rensselaer Polytechnic Institute (RPI) set out to develop an interdisciplinary mechatronic system by designing and prototyping a two-wheeled robotic locomotion platform inspired by (and with the permission of) the Segway Corporation, maker of the Segway Human Transporter. The two-wheeled, self-balancing transport platform utilizes parallel-wheel locomotion to provide precise maneuverability while maintaining system stability. The team tackled both the complexity involved in modeling, analyzing, and controlling the platform, as well as the implementation of two fully operational prototypes in a four-month time period.
Professor Kevin Craig, the course instructor and mentor, emphasized the synergistic combination of electrical engineering, mechanical engineering, control systems, and computers in his Mechatronic System Design course. The team followed the Engineering System Investigation Process (see Figure 1) to model, analyze, validate, and then design their control system. They prototyped the LOT-V (Light Object Transport Vehicle – see Figure 2), which allowed sensors, actuators, and control implementations to be tested on a lightweight, tethered system. Once satisfied with the prototype, they developed, using all they had learned, a rugged, self-contained platform named the HOT-V (Human / Object Transport Vehicle), capable of transporting an adult person.
Figure 2. LOT-V balancing with tether.
National Instruments hardware and software, functioning in tandem, was included at every stage of the project. From initial model development and parameter identification, through model simulation and validation, to design prototyping and final deployment, it was crucial in the development of both the LOT-V and HOT-V platforms. The synergy between hardware and software empowered the team to navigate the design process using a single software environment which allowed simulation, design, and prototyping to intertwine seamlessly. The team utilized LabVIEW with the LabVIEW Simulation Module not only in system simulation, but also in prototyping and deployment on multiple targets including the CompactRIO.
Identify, model, and design
The first step in development began with the Engineering System Investigation Process, which is the process used to understand how a system actually works and how it can be improved. This process started with the development of a physical model of the system by making several simplifying assumptions. Physical model parameters were identified and mathematical equations were derived by applying the laws of nature to the physical model.
Computer simulations allowed the team to explore and understand the equations of motion by analyzing the system in different configurations. Modeling, analyzing and simulating components of the system allowed the team to correctly choose and size components as well as test control schemes without resorting to a trial-and-error approach.
Prototyping the LOT-V
After a careful analysis of the simulation results, it was decided that two variations of the platform would be created. The first, LOT-V, was a small, sensor-rich prototype that was used to explore and evaluate different sensor solutions and control algorithms. Its size allowed testing and exploration with fewer safety concerns. It was built using off-the-shelf parts and controlled with an inexpensive NI-DAQ PCMCIA Card tethered to the robot.
The second, HOT-V, was scaled specifically to transport an adult human. It incorporated the lessons learned from the LOT-V prototype, in both mechanical and electrical design, to be as safe and modular as possible. The HOT-V is controlled with the NI CompactRIO control platform. The CompactRIO provides a rugged, high-performance real-time computation package coupled with high-speed, electrically-isolated I/O, making it an ideal self-contained control system.
Both prototypes utilized NI LabVIEW and the LabVIEW Simulation Module for algorithm development and deployment. LabVIEW was absolutely essential to the success of the project due to the ease and speed with which the software could be reconfigured and shared between platforms. Algorithms and software architecture used on the LOT-V were effortlessly ported over to the HOT-V for execution on the CompactRIO with minimal modification.
The LOT-V’s primary purpose was to act as a test bed for the various types of sensors that we planned to use on the project. We devoted a significant amount of time to choosing appropriate and reliable sensors because accurate and high-bandwidth information was necessary to successfully balance the platform. One additional challenge in choosing the correct sensor was sensor placement. We planned to attach all sensors within the platform structure increasing the types of terrain the platform can operate on. We quickly found that several different sensors would have to be integrated together to make a single measurement of acceptable quality. Testing sensor combinations was extremely easy in LabVIEW because the entire system was implemented in a single environment. Schemes could first be evaluated in simulation and then transitioned to real-world hardware with no extra effort.
Figure 3. The LabVIEW simulation subsystem for the Doebelin sensor fusion scheme.
Many sensor schemes were explored and after extensive experimentation, a sensor fusion scheme introduced by Professor Ernest Doebelin of Ohio State University was chosen (see Figure 3). It was then integrated into the overall control system of the platform and used to combine measurements from an inclinometer and a rate-gyroscope to calculate the angular position relative to gravity (see Figure 4).
Figure 4. Sensor data flow overview.
The final implementation also used motor-mounted optical encoders to accurately steer and to also measure the linear position and velocity of the platform.
The team implemented a full-state-feedback controller to control the balancing of the platforms using the LabVIEW Simulation Module. It was developed using the Linear-Quadratic Regulator (LQR) technique on a linearized version of the system (see Figure 5). The flexibility of the NI software-hardware platform allowed different variations of the controller to be easily tested to determine which would be the optimal choice for the actual system.
Figure 5. Lot-V NI-DAQ-based control within LabVIEW simulation loop.
Once the team was comfortable with the overall design, it was time to use the models, sensors, actuators, and control schemes they had prototyped to build and deploy a rugged, independent
Deployment of the HOT-V
Figure 6. HOT-V balancing without a passenger.
While the HOT-V (see Figure 6) was a much larger and more complex platform, it utilizes everything the team developed and learned from the smaller LOT-V. It is constructed from primarily off-the-shelf components and went from an assortment of parts to a completely integrated and operational platform in just a few days. Once assembled, the software was amazingly simple to port from one LabVIEW target to another. They were able to take the complete control system for the LOT-V, compute new controller gains, modify the I/O to work with the CompactRIO platform, and deploy it in a few hours (see Figure 7).
Figure 7. HOT-V real-time control within the LabVIEW simulation loop.
The HOT-V platform, in addition to being much larger than the LOT-V, is also much more powerful (see Figure 8). It has a power support system capable of delivering several hundred amperes of current allowing the motors to run at over 3 horsepower. CompactRIO was chosen for its shielding and isolation properties, in addition to its small, rugged form.
Figure 8. HOT-V platform components.
While the HOT-V platform was totally self contained, it was important for us to be able to monitor and interact with the system during operation. The internet-ready LAN adapter on the CompactRIO allowed us to install an 802.11 wireless router inside the system. Thisdecision increased the functionality of the system by an order of magnitude by creating a wireless communication connection with the system; the team was able to control the HOT-V from multiple platforms.
The team then decided to integrate an HP iPAQ Pocket PC into the operation of the HOT-V, using the LabVIEW PDA Module to stay on the LabVIEW platform. The Pocket PC runs a LabVIEW program that allows two operating modes:
Heads-Up Display: While the user is riding the HOT-V, the iPAQ displays information about the state of the platform, such as how much battery life is left and how much power it is currently consuming. (The iPAQ can be clipped on to the HOT-V’s handlebars.)
Remote Control: When the user is not riding the HOT-V, the user can remotely control the platform from the iPAQ. The user has the option to control various parts of the system and control movement.
Adding the palm-sized user interface turned the HOT-V into much more versatile experimental platform.
While the team was able to conquer many significant design and implementation challenges, perhaps the most significant was the short time frame in which it was undertaken and completed. The entire design from conception, through modeling and analysis, experimental validation, and control design, to construction and implementation was completed in a four-month period. The team attributes this success to the single development platform and tight integration of the NI Hardware and NI LabVIEW software that allowed for a streamlined progression from modeling and analysis, to prototyping and deployment.
Rensselaer Polytechnic Institute