Top Three Challenges in Robotics

Publish Date: Jul 27, 2017 | 5 Ratings | 4.40 out of 5 | Print | Submit your review

Table of Contents

  1. Better Batteries
  2. Better Actuators
  3. Industry-Grade Robotics Software
  4. Software In-Work
  5. What’s Next for NI?

After meeting with leaders in autonomous system design at a few of the top robot design shows (Association for Unmanned Vehicle Systems International (AUVSI) and NIWeek 2009 Robotics Summit), I wanted to share the most impeding challenges in this industry according to top roboticists. By sharing this information, roboticists can work together to solve these issues and catapult the robotics market forward to the $470 billion dollar industry it is predicted to be in 2020, enabling innovation that will truly improve everyday life.

Recently, one of the most well-known roboticists in the world spoke at the annual NIWeek graphical system design conference. Dr. David Barrett, current director of the SCOPE program at Olin College and former vice president of iRobot and director of Disney Imagineering, summed up our three biggest challenges:

• Creating better (smaller, lighter, and more powerful) batteries
• Creating better (smaller, lighter, and more powerful) actuators
• Addressing the need for an industrial-grade, hardened, and richly supported software development system

These challenges were echoed by other leaders in other forums. Dean Kamen, founder of FIRST robotics and DEKA Research and Development Corporation, expressed his frustration with the traditional embedded approach for robot design and described why his team chose NI LabVIEW software and NI CompactRIO programmable automation controllers (PACs) for intuitive and powerful control of each autonomous robot for the FIRST program. Ellen Purdy, Department of Defense director of ground robotics and autonomous systems, focused on the need for more standardization and more endurance (battery power) in our autonomous systems. Dr. Hee Chang Moon from Pohang Institute of Intelligent Robotics emphasized the need for an integrated embedded system such as CompactRIO that works with intuitive software, and Michael Fleming of TORC Technologies explained why TORC switched from using legacy programming tools to LabVIEW to design its sophisticated autonomous systems. He specifically spoke to how TORC benefits from the high-level yet powerful, software that helps abstract the complexities of the system design.



Figure 1. Dean Kamen of FIRST Robotics says that even high school students can address some of the most complex robotics needs with tools like LabVIEW and CompactRIO.


In hearing these challenges from design engineers, robotics experts, professors, and students from around the world, I wanted to share some insight into what National Instruments is developing. We have heard you loud and clear – you need help to design better robots and autonomous systems and you want LabVIEW and reconfigurable I/O technology to help you. Here’s a sneak peek into what we’re working on.

1. Better Batteries

Although NI does not currently make batteries, we are planning to add some features to our embedded hardware to help with this challenge. We have added new capabilities to our RIO hardware road maps specifically for autonomous system designers. We are working to add built-in self-diagnostic and monitoring APIs to CompactRIO and NI Single-Board RIO platforms. You can use these functions to gain greater insight into your power status and to design more graceful shutdown and power-save modes.

We are also are working to greatly increase the accessibility and optimization around energy usage and alternative forms of energy. From wave harvesting to solar and wind power, these worlds of green engineering and robotics might soon join to offer viable solutions for both, including extremely efficient power generation and harvesting on small mobile platforms. We are also working to provide more efficient control algorithms like field-oriented control (FOC) for field-programmable gate array (FPGA)-based control. FOC, also known as vector control, improves sinusoidal control by providing high efficiency at faster motor speeds. It delivers the highest torque per watt of power of all the control techniques. State-of-the-art technology mergers like advanced control and green engineering would make it possible to develop numerous critical applications, from wave-harvesting powered surveillance systems in our oceans to solar-powered agricultural machinery to empower our farmers to be more efficient.


Figure 2. FOC algorithms with the LabVIEW FPGA Module can transform an existing motor control system into a high-performance, high-efficiency machine or robot.

Back to Top

2. Better Actuators

As with batteries, NI does not plan to make actuators, but we do have more than 25 years of experience with actuator connectivity as well as a current focus to improve the interaction with and control of actuators. We are working with leaders in the industry such as Maxon Motor to make programming these devices much more straightforward. Our design team has firsthand experience with the challenges of selecting and programming actuators. For example, we recently built a new demonstration platform called VINI with the goal of creating an autonomous simultaneous localization and mapping (SLAM)-capable platform using a Hokuyo LIDAR, Ocean Server compass, the CompactRIO embedded control platform, and the new NI industrial controller for vision processing with an NI Smart Camera and a low-cost Ethernet camera from Axis. We wanted to implement a mechanam drive system with CANopen-based brushless servo drives from Maxon Motor. Moving through the design process, we quickly discovered the importance of proper motor selection, particularly with different locomotion options like Mecanum or omnidirectional wheels. You must compensate for different drive and torque needs than you would require with with a standard four-wheel drive system.


Figure 3. This shows the VINI robot architecture and physical robot, complete with CompactRIO and an NI industrial controller for reliable sensing, thinking, and acting.

As mobility trends continue to move from traditional four-wheel systems to novel biomimetic mechanisms that replicate nature, the need for sophisticated actuators grows tremendously. For example, the July 2009 feature article in Robotics Business Review described a project called RoboSwift that mimics the form of a swift or swallow bird for reduced probability of detection and better maneuverability in urban areas. RoboSwift’s wings even have a variable sweep, so the flying robot can change wing geometry for better turn radius. Designs like these demand actuator innovation.

NI is helping by working diligently to make motor control and interfacing easier. From motor drives like our NI 9505 and the recently released NI 951x family of drive interface modules for CompactRIO to intuitive motion control abstraction in the LabVIEW 2009 NI SoftMotion Module, LabVIEW motor control examples (ni.com/ipnet) for FPGA-based controller area network (CAN) communication, and simple digital control like PWM, we continue to make actuator control easier, more efficient, and highly flexible.

Back to Top

3. Industry-Grade Robotics Software

The biggest challenge expressed by experts is the one I am most excited about. Roboticists need powerful software to design their autonomous systems – software that is not unique to a particular robot or task, is open to incorporate existing algorithms, and is powerful enough to solve problems we do not even understand today. This is the challenge I see being met the soonest. With your input and some of the leading software developers in the world, we can empower designers with a powerful software package within a year’s time. The most opposing force is mind-set. Many companies are invested in taking the wrong approach. AnneMarie Bourcier of Aldebaran Robotics in Paris, France was recently quoted on Slashdot.org saying, “It’s easier to build everything from the ground up right now.”

This is the mind-set of the last decade in robotics and the reason we have not seen massive adoption or progress in our field. Instead, we need to see some convergence of technology and enable the innovation we need to make the kind of impact that robots can. Thus, robotics needs standard, industry-grade software.

1. This software must be intuitive.

Many roboticists have a mechanical engineering or electrical engineering background and do not have the time or money to learn the ins and outs of the most useful computer science techniques, but they do need to be able to take advantage of those techniques. Programming capabilities like object-oriented programming and recursion can be critical in a robust autonomous system, so they need a language that is capable of these features. Additionally, the user interface must be intuitive and flexible. Often the end users of the autonomous system are rescue professionals like firefighters, soldiers, or the elderly, none of whom should be faced with a complex or confusing interface to complete their task.

2. The software must integrate well with I/O.

Every autonomous system must sense or perceive the world around it as well as act on that environment. Sensing requires input from external sensors such as laser rangefinders and sonar sensors, and acting requires the ability to drive many different types of actuators. In addition to sensing and acting, your software tool must easily implement your application on real hardware, meaning it must have strong integration with real-time OSs, real-time embedded hardware, and even FPGA-based devices. Many software packages, including Microsoft Robotics Studio, on the market lack this capability – they can simulate and run on your development machines but fail to provide any kind of real-time implementation capability.

3. The software must be open and flexible.

Many autonomous algorithms have already been optimized and are ready to be reused, and the building blocks of others have been created but need to be built on for greater capabilities. For example, many systems start with basic search algorithms such as A* and D*, which you can find in numerous places. Many roboticists wish to start from those basic algorithms and add their own innovation or latest research on top of those building blocks to create a new type of search or integrate them with a new mapping technique to aid robot rescue or medical assist applications. The closed nature of many robotics software packages today frustrates design engineers. Packages like iRobot Aware work fine for their robots but cannot be used for customization or for your own unique robotic design.

4. The software must be interactive.

Innovative robot design is not straightforward and requires many iterations and prototypes. The software used to design the robot should be able to accommodate and enhance this experience. Roboticists need a software package that they can easily debug, that integrates intuitive simulation, and can quickly be implemented on a real-time hardware system to test the algorithms and design with real-world I/O. That same code then needs to easily port back to the design environment for additional optimizations or software tweaks.

Back to Top

4. Software In-Work

The NI R&D and product development groups are working hard to address these challenges and provide an intuitive, I/O-integrated, open, flexible, and interactive software environment for building autonomous systems. We have a talented team with decades of software design expertise working to design a version of LabVIEW to address these needs. LabVIEW 2009 meets these requirements. In fact, Dr. Barrett claims that LabVIEW is the software the industry needs, describing its ability “to deal with webs of sensors, multiple actuators, and complex dynamic control algorithms that can be easily implemented in real time.”

Our design team is working to create new functionality including the integration of robot-specific sensors and actuators, obstacle avoidance, and search and kinematics algorithms, as well as new visualization capabilities. We strongly believe that you should not need a doctorate in robotics to build robots. You should be able to benefit from powerful design tools and develop sophisticated robots with elegant simplicity.



Figure 4. Robotics software design does not have to be complex. This simple LabVIEW diagram provides LIDAR sensing, obstacle avoidance thinking, and skid steer acting capabilities with intuitive and readable programming.

Back to Top

5. What’s Next for NI?

Stay tuned to ni.com/robotics for the latest news and you just might stumble upon something new before the end of the year. Also, check out the Robotics Fundamentals Series to learn more about using LabVIEW and NI hardware to develop your robotics application. And never forget…robots rock!

Shelley Gretlein

Shelley Gretlein is the Real-Time and Embedded Senior Group Manager for National Instruments. Currently focused on robotics and autonomous system design, she is responsible for the development strategy and worldwide promotion of the LabVIEW Embedded Platform including LabVIEW Real-Time, LabVIEW FPGA, and LabVIEW Microprocessor products, as well as system design partnerships and technology awareness. Actively involved in industry consortia on industrial control and embedded forums, Gretlein has been a graphical system and embedded design consultant for various publications. She joined National Instruments in 2000 and holds a Bachelor of Science in Computer Science and Management Systems from the University of Missouri-Rolla.

Back to Top

Bookmark & Share


Ratings

Rate this document

Answered Your Question?
Yes No

Submit