Robot drivers take the drudgery out of testing

Virtual testing may be the all the rage, but there are still many tests that must be run on physical prototypes. For example, you can't perform a proper emissions test unless you run an actual vehicle through a series of starts, stops, and speeds while you collect and analyze the emissions gases.

Until recently, human drivers were the ones to put the cars through their paces. While human drivers are able to do this job well, they are prone to making mistakes. A human driver could easily miss a shift or an acceleration, which could invalidate the test and require that you start it again from scratch.

Another problem with using human drivers is repeatability. A person will never drive through a cycle in exactly the same way twice. For some tests, you can mitigate this variation by averaging the results of many different runs of a particular test. In other cases, however, running multiple tests is too difficult or too expensive. Imagine that you need to compare the test results of two different designs. If your tests were repeatable, you would save money by running just a small number of them and comparing the results.

Enter the robot driver. Robot drivers take the drudgery out of testing by replacing human drivers and freeing them up to do more productive and creative tasks. And as an added benefit, they perform more repeatable tests as well.

Robot drivers come in a variety of configurations. The Stahle SAP2000 (Figure 1) sits in the driver's seat and connects to the vehicle's accelerator and brake pedals, as well as to the clutch and shift lever if the vehicle has a manual transmission.

The Anthony Best SR series (Figure 2) is used for steering control only. These robots connect to the steering wheel and can automatically guide the vehicle through a preprogrammed series of maneuvers, such as J-turn maneuvers and rollover tests.

You can also find robots that only perform braking. Such robots can be programmed to apply forces to the brake pedal for a specified period.

The use of the braking robot is a good example of the repeatability advantages that robots enjoy over human drivers. A European auto company compared the performance of a braking robot and of experienced human test drivers during an emergency braking test.

The test profile was simple. The driver had to apply a force of 400 N to a vehicle's brake pedal in order to decelerate the test vehicle from 160 kph (a little more than 80 mph) to 0 kph. The test specifications also stated that the force on the pedal was not to exceed 500 N or drop below 300 N during the test.

The human driver was able to meet these requirements in only three of the 27 trials. The braking robot, on the other hand performed the test successfully in five consecutive tries, with the force on the brake pedal varying from 400 N by only 30 N maximum. Instead of taking an entire day to produce an acceptable number of test runs, the company was able to do it in only a few test runs with the braking robot.

Since a robot driver is just like any another piece of test equipment, you should figure out what types of tests you're going to run and what vehicles you need it to control before purchasing one. Brian Bucalo, manager of Stahle Robots and the test cell products division of Accurate Technologies (Wixom, MI; www.accuratetechnologies.com), says the most common mistake buyers make is purchasing the wrong type of robot. He notes, "We have seen customers spend a lot of time after a purchase figuring out how to use the robot. There are so many options available that good decisions up front will save everybody a lot of time and frustration."

Phil DeBerry, a representative for Anthony Best Dynamics (Milford, MI; www.abd.uk.com), says that the most important specifications to consider when buying a robot driver are:

• performance (torque/speed/time capability),
• software features (tests available, ease of setup),
• physical design (mass and inertia of mechanical components as well as the drive used for the actuators),
• ease of installation,
• measurement capability (angle/torque, external data capture), and
• safety considerations.

Not understanding how a robot driver's software works is a mistake that new users often make. DeBerry says, "The most common problems that our new users have come from their lack of experience with our software and its full capabilities." DeBerry recommends that users, especially new users, try to build some time into a schedule to allow the test engineers to come up to speed. Having some time to learn the system software will help system integration go more smoothly and will allow you to take advantage of all of the robot's features.

For example, most of the robot drivers have a "learn" function that allows you to teach the robot a particular driving sequence or maneuver. Spending some time to learn how this works should help you train a robot more quickly, and less trial and error in teaching a robot new tricks should help you save test-development time.

You will also have to spend some time learning how to interface the robot to the test system's host computer. The Stahle robots, for example, have three different test interfaces. The first is the stand-alone mode. In this mode, the robot basically controls itself and has little or no interface with the host computer.

The second type of interface uses a table of states set up by a test engineer and preprogrammed into the robot. The host computer issues commands to the robot controller, which then cycles through the states upon command. Third, the host computer can take full control over the robot system, using analog signals to drive each actuator directory.

Deciding which control scheme to use requires careful consideration. For example, when using a robot driver during a high-performance test, you may want to choose direct control so that the time required to actually implement a control command is as short as possible. When running an emissions test, however, this may not be as much of a concern, and using the second type of control interface will suffice.

Whatever method you choose to control the driving robot, DeBerry left me with this last bit of advice: After development of any maneuver, either on the test track using the robot's teach function or graphically using the GUI, always, always try out new tests at a low vehicle speed. Ensuring that it works right at low speeds will help ensure that it will work reliably and safely at high speeds.

Here are examples of robot drivers that are currently on the market.

The Steering Robot from Anthony Best Dynamics connects to a vehicle's steering system for testing its transient handling behavior on the test track. The Steering Robot allows you to apply a wide range of steering inputs with high precision and repeatability and to gather high-quality data quickly. You can also use it to test a vehicle's steering system.

The Steering Robot is available in several versions with differing motor performance. The basic unit has a peak torque capability of 30 Nm, but the manufacturer can supply robots with peak torque delivery of over 150 Nm if required.

The standard Steering Robot software has many built-in test profiles that reduce setup time. These tests include sinusoids, J-turn and fishhook maneuvers, rollover tests, and power-steering characterization tests. A Path Following option enables the Steering Robot to accurately steer a vehicle through a defined path.

Anthony Best Dynamics, Wiltshire, England. +44-0-1225-860-200; www.abd.uk.com .

The ADS-7000 Automatic Driving System automatically operates a vehicle on a chassis dynamometer, using an accurate servo-tracking mechanism. Its repeatability and accuracy exceed capabilities of the most experienced human drivers.

Designed for vehicle-emission, fuel-consumption, and emission-endurance testing, the ADS-7000 drives nearly any vehicle type, including CVT, LEV, direct-injection, hybrid, and electric vehicles. It can operate both manual-transmission and automatic-transmission vehicles, and it automates tests conducted in harsh or extreme conditions.

A sophisticated "learning function" automatically determines the positional relationship between each actuator and the vehicle. Once data is stored, the ADS-7000 can test vehicles of the same model without performing the learning function.

The PC-based control system provides overall test sequence control, real-time graphic displays, and editors for entering vehicle model data, driving schedules, gearshifts, ignition schedule, and other test parameters.

Horiba Instruments, Ann Arbor, MI. 800-346-7422; www.horiba.com .

The SAP2000 operates in environments ranging from —40°C to +80°C with a high degree of repeatability. Over a 17.86-km emission driving cycle, for example, the SAP2000 will typically come within 1 m of the overall distance.

Installation is quick and easy. The system sits in the driver's seat and connects to the vehicle's pedals in less than 10 min. The robot can learn a driving cycle and operate independently or it can connect to a test bed's host computer system for control by the test bed.

The SAP2000 is available in a variety of configurations, some of which include stand-alone actuators that enable you to run tests on transmission or engine dynamometers similar to those run on chassis dynos. This feature allows you to correlate component tests with vehicle tests. You can also use the SAP2000 robot with steering robots to fully automate proving-grounds road tests, or tests on flat-belt test beds.

Stahle, Steinegg, Germany. +49-0-7234-8570; www.stahle.com .