Researchers develop prototype vehicle for blind drivers

Researchers from Virginia Polytechnic Institute and State University have developed a prototype semiautonomous vehicle that enables a blind driver to navigate, control speed, and avoid collision through a secure driving course. This has been done to promote the often underestimated capabilities of the blind and to inspire innovation in the development of blind access technologies, with the project being laid down as a challenge by the USA's National Federation of the Blind. The Robotics and Mechanisms Laboratory (RoMeLa) at Virginia Tech believes it was the only organisation to accept the challenge.

Re-established in 2008 as a senior design team and undergraduate research project within the Department of Mechanical Engineering, the Virginia Tech Blind Driver Challenge (BDC) defined the initial goals for the world's first working prototype of a blind driver vehicle. In only two semesters and with nine undergraduate students and $3000 in seed funding, a blind driver would be expected to safely perform three fundamental driving tasks: navigate through a curved driving course defined by a single lane of traffic cones, regulate speed within a predefined limit, and exhibit sufficient emergency-stop capability to avoid colliding with an obstacle.

Prototyping platform

NI products have been used as the singular hardware and software interface for the blind driver system since the project's inception. NI products were chosen because the team needed a cost-effective prototyping platform, short data acquisition and processing time to minimise lag in time-critical driving environments, compatibility with numerous sensors and devices, power and reliability in demanding testing conditions, an intuitive programming interface, modularity, size, weight, and capacity for hardware expansion during future development. The researchers examined RoMeLa's long history of success in using NI products in a variety of applications, from humanoid soccer-playing robots to fully autonomous vehicles. These applications, in addition to the blind driver system, serve as a testament to the versatility and ideal functionality of NI hardware and software as a prototyping platform for robotics applications.

Environmental perception

The current blind driver system consists of various sensors and novel non-visual driver interfaces attached as a modular system to a modified dune buggy. A Hokuyo UTM-30LX single-plane laser rangefinder (LRF) was used for environmental perception to scan the driving environment for cones and other obstacles and feed that information to an onboard CompactRIO real-time controller and its real-time field-programmable gate array (FPGA) processing targets. Conveniently, existing NI device drivers support Hokuyo LRF products because NI engineers provided a custom driver before the UTM-30LX was released to the general public.

A laptop running LabVIEW software provides temporary USB hosting capability to the CompactRIO controller. The laptop also allows a sighted passenger to passively monitor the operation of all hardware and software and easily modify any heuristic-based programming for quick calibration during field testing.

Additional sensors gather important information regarding the state of the vehicle, such as speed from a Hall effect sensor and steering angle from a string potentiometer. Data is acquired from these sensors and processed directly using the high-speed FPGA on the CompactRIO real-time controller.

Driver interfaces

After collecting an image of the driving environment using the various sensors, the information is processed and transmitted to the driver through non-visual cues. The ultimate goal when developing a non-visual driver interface (NVDI) is to effectively and efficiently provide information to a driver to maximise situational awareness and allow the driver to make quick and precise driving decisions. The array of NVDIs on the first iteration of the vehicle is a combination of informational and instructional cues for safety and redundancy.

For speed regulation, the driver can operate at a comfortable speed until reaching a maximum speed limit, at which point a vibrotactile vest on the seat belt informs the driver what degree of braking is necessary to return to a safe operating speed. If the vehicle detects an unavoidable collision with an obstacle, the vest cues the driver to stop the vehicle immediately.

During initial vest testing, a custom circuit board was used to control the motors. RS232 signals were sent from the LabVIEW software on a PC to a PIC microcontroller, which controlled a large bank of transistors and relays to actuate motors in the vest at various intensities. After acquiring CompactRIO, the circuit board was no longer required because of the NI 9485 eight-channel relay module. Bypassing the circuit board reduced the bulk and potential complications from additional hardware, which greatly simplified the underlying software, and significantly reduced the time between obstacle detection and full motor vibration, which is critical for drivers in an emergency situation.

For steering guidance, a potential field algorithm provides the path generation. After calculating a path, the system instructs the driver where to steer to stay in the lane and avoid obstacles. The driver is told how many 'clicks' to turn the steering wheel via a pair of headphones and LabVIEW text-to-speech software. A mechanism attached to the steering column clicks every five degrees to provide precise audible feedback.

Additionally, a prototype tactile map was developed, which is conceptually similar to a high-resolution grid of regenerative Braille. The map places an image of the surrounding environment literally in the hands of the driver. Similar to the tiny holes on an air hockey table, a physical map is generated by passing compressed air through small 'pixels' to depict the surrounding obstacles detected by the laser range finder. This device, appropriately named AirPix, allows the driver to 'see' the surroundings and navigate safely through them. The audio and vibrotactile NVDIs are still necessary for redundancy, but using the driver's high-bandwidth sense of tactation through this tactile map technology makes data pathways available for other driving uses, such as listening and interacting with a GPS through voice-recognition software for higher-level path planning.

Benefits of NI hardware and software

Using NI hardware and software from NI, the team created what is believed to be the world's first working prototype of a blind driver vehicle. They state: "With limited funding and development time, NI products played a vital role in the success of the project by providing an easy-to-use and cost-effective prototyping platform. The intuitive graphical programming interface of LabVIEW made it easy for a team of undergraduate mechanical engineering students to quickly and efficiently create custom embedded software without the need for expertise in specific text-based programming language.

"The modular design and large capacity of CompactRIO for additional I/O modules combined with the extensive compatibility of LabVIEW with external devices ensured that future expansion and improvements to the system would be possible with minimal effort or cost. The real-time and FPGA processing targets provided the high-speed data acquisition and processing power necessary to gather essential data from the time-critical driving environment. In addition to the internal capabilities, the convenient size and low weight of CompactRIO were ideal for the limited space and payload capacity on the current blind driver buggy.

"Throughout the highly iterative prototyping process, NI modular products made it easy to adapt to unique and demanding testing environments, changing vehicle platforms, and shifting project objectives. With versatile NI hardware and software, the Virginia Tech Blind Driver Challenge continues to 'invent the future' in blind access technology." [The team included Dr Dennis Hong, Greg Jannaman and Kimberly Wenger of the Virginia Polytechnic Institute and State University.]

For more information about NI CompactRIO hardware and LabVIEW software go to www.ni.com.