Checklist for robotic survival

Roboticists who ignore one or more of three principles when designing robots will fail, and many do. Six systems require attention for mobile robotics.

By Mark T. Hoske March 1, 2013

Robot designers who ignore one or more of three principles are doomed to fail, according to a roboticist with 22 patents who has worked for three robot companies and MIT Robotic Lab. Joseph L. Jones, co-founder and chief technology officer, Harvest Automation, told attendees at the Robotics Industries Forum about a checklist for robot survival.

“How are robots like sea turtles? For each 1,000 sea turtles hatched, only one lasts to adulthood. It’s about the same statistic for robots. Robot designers need to be really careful in choosing which robots to build,” Jones said. He also worked at iRobot (which has sold 9 million Roomba robotic vacuums) and at Denning Mobile Robotics. In 30 years of watching the robotic industry, he’s observed three key elements that need to be present with any robot project for success.

Checklist for robot survival

To create a successful robot, Jones said the robot should:

  1. Do something that lots of people want done
  2. Be built with existing technology, and
  3. Be cost-competitive with current solutions.

While this seems perfectly logical, when in the thick of things, roboticists usually ignore one and often all three, Jones said, in his 30 years of observations. In his talk, “Small Mobile Robots for Agriculture,” Jones discussed his current efforts at Harvest Automation, founded in 2007, now with 40 employees. He’s making robots for agriculture. A nursery and greenhouse (N&G) robot is the company’s first product.

Plants sold in garden stores often are grown in pots in open fields that extend to the horizon. Armies of hard-to-find workers manually space pots in the field after the pots are unloaded from wagons.

“After I observed that, I thought, ‘If we cannot build a cost-effective robot to do this, we’re in the wrong business,’” Jones said.

Six systems challenge mobile robots

With any mobile robotic design, there are six system challenge areas: application system (sensors, actuators, and software), navigation, hazards, mobility, power, and the interface. (For the HV-100 the application system consists of the gripper, the laser ranger that identifies the pots, and the software associated with pickup and putdown, he said.)

Application notes: N&G work involves a lot of pain, with repetitive heavy lifting, Jones said; required seasonal labor is scarce and 80% undocumented. It’s also inefficient, Jones noted. For this project, no research grants were needed. Picking up the pot required a one-degree-of-freedom manipulator. No GPS or cameras are used. A sensor consisting of a pair of photo diodes is used to find the yellow tape border guide; an off-the-shelf range sensor locates the pots. No inter-robot communication is needed to grab a pot and add it to the pattern in order. Batteries require a swap in 3-5 hours, depending on plant weight and terrain conditions.

Three years of development has led to the delivery of the first products to a customer in Georgia. A robot supervisor adjusts the user interface when needed and moves the boundary marker. Beyond placement, future capabilities will accommodate plant maintenance, putting the pot into a booth for grading, pruning, and targeted application of insecticide or nutrients (avoiding risk of human contact with whirling blades or hazardous chemicals).

Seek a large, neglected market

“We kept the design as small and simple as possible and tried to find a large and neglected market,” Jones said. Several companies have proposed orange-picking robots, but that would be higher complexity and just a $2.5 billion market compared to N&G at $17 billion, he cited.

In general, when considering a robot, it’s often helpful to reimagine the task. Consider the numerous design differences between an often vertical human-operated vacuum and a Roomba vacuuming robot from iRobot, which looks more like a large hockey puck with one button on top.

How can applications be found for robots? Roboticists usually don’t know specific industry details. Potential customers usually don’t know robots can help. To find opportunities, brainstorm, be opportunistic, and visit promising sites. Evaluate the applications with large market sizes and customer needs, simple technologies, and low cost.

Good examples include the Roomba, with 9 million units sold; the Kiva warehouse robot; and the Aethon hospital tug that pulls cabinets around for nurses, saving valuable staff time and effort.

Reimagining agriculture

There will be other agricultural applications with the anticipated global population increases. From now to 2050 we have to increase food production by 70% without more land or more water than used today. To increase yield, consider the giant vegetables displayed at county fairs. They get to be giant-sized because of lavish attention given to them. Robots may help optimize crops, supplying the missing labor. Selective pruning and hexagonal planting, impossible with traditional tractor wheels, may increase yield by 15% for some crops. Polyculture crops (multiple species in one field) can reduce pests and improve soil use, with fewer pesticides. Use of robots also may permit multiple harvests, taking fruit only at its peak and allowing immature fruit to continue to ripen. Robots could optimize crop growth; minimize and localize pesticide, water, and nutrient use; work safely with farmers; and avoid harmful soil compaction. (Compact soil increases water runoff and eliminates aeration.)

First model, Harvest Automation HV100, costs about $30,000, with return on investment after 12-18 months of use, Jones said.

Robots are powering the next innovation surge in agriculture, Jones said. Harvest Automation, backed by a team of robotics innovators, has engineered the first practical, scalable robots for a range of agricultural applications starting with nursery and greenhouse operators, Jones explained. Addressing labor scarcity issues, Harvest’s robots are designed to work alongside people, not replace them, in a grower’s operation, to create a sustainable workforce combining robots and people to increase efficiency, reliability, and plant quality, he noted.

– Mark T. Hoske, content manager, CFE Media, Control Engineering and Plant Engineering, mhoske@cfemedia.com.

www.harvestai.com 

www.robotics.org – Robotic Industries Association


Author Bio: Mark Hoske has been Control Engineering editor/content manager since 1994 and in a leadership role since 1999, covering all major areas: control systems, networking and information systems, control equipment and energy, and system integration, everything that comprises or facilitates the control loop. He has been writing about technology since 1987, writing professionally since 1982, and has a Bachelor of Science in Journalism degree from UW-Madison.