This dissertation, and the research contained within, was carried out to reduce the side collision in school and other buses that ply our roads. The most significant of these problems is in road transportation. Research and development related to side collision warning systems have been directed at lightvehicles and long-haul trucks. Little or no research has been directed towards the usage of these in transit buses. These are necessary for transit buses too in a different manner - to detectpedestrians and other vehicles that normally miss the eye of the driver appearing in his blind spot. Therefore, the focus has been to crticially analyze the requirements of transit buses in terms of reducing collision through a warning system that can effectively reduce accidents and mishaps on our roads today. Transit buses are widely used in many countries today. The developed countries make use of it mainly for transporting children to schools and back, however are more widely used in developing countries and nations where distances, that need to be travelled are not too high.
This analysis discusses user-interaction and interface with these devices and the warning system, while providing a holistic approach towards understanding how they will operate. Subsequently, throughout the duraction of this research it was proven that there is a need to improve the sytems itself by incorporating modern technology in a control panel format and adapt it to the needs of the driver's cockpit environment. This led to the observation that the driver's area could effectively become the nerver centre of the transit vehicle. The result of this research was a design brief that highlights the need for a product which allows a user-friendly method of a collision warnign system, and all its linked electronic resources. Thus, by raising a comprehensive and inclusive awareness of all the relevant issues, the dissertation has provided the basis for such a product to be successfully developed.
The following document describes research done in the dissertation, and defines a design brief for an innovative product design. Only information relevant to driving to the design outcome is included.
The document sets out with a carefully considered topic. Limitations and boundaries are placed on this topic in order to help control the investigation stages of this dissertation. Reasons for choice in topic are also given along with a brief description of the process, which was used to arrive at this decision. The journey will outline the events encountered that lead to the final topic and outline some of the concepts that were had.
Following this is an overview of the methods used to arrive at the design outcome. This covers a description of how the study was planned and what was determined to be relevant information to gather for the study, how this information was gathered., how informated was evaluated in terms of reliability and its value towards the design outcome, and finally a discussion of the techniques that were used for information retrieval and analysis.
A presentation of the information gathered and relevant towards the desing outcome or design brief is then provided. This information is given, as factually as possible with no opinions or beliefs included; an analysis follows that includes a critical evaluation of this information.
A design brief supported by the information preceding it is then proposed. This design brief outlines in as much detail as possible for the functionality and the requirements for the design concept to be developed. Additionally, a discussion of possible futures for the product is illustrated, details of considerations for features that could be included.
Glossary of terms
Accessible- A term used to indicate a vehicle, vessel, facility (including bus stop or terminal) can be used, entered into, and reached by persons with disabilities.
Adult- A passenger between the ages 19 and 64.
APTA- American Public Transit Association, a non-profit organization which serves as an advocate for the advancement of public transportation programs and initiatives in the United States.
Arrival time- The time a transit vehicle reaches a bus stop or destination.
Articulated bus- An extra-long, high capacity bus that has the rear body section flexible but permanently connected to the forward section.
Bus- A rubber-tired transit vehicle designed for roadway operation to transport a large number of persons for public transportation service. In most cases, it operates with a self-contained source of motive power.
Bus driver or operator- The driver of a bus transporting persons for public transportation service.
Bus pad- A bus stop located along certain highway exit ramps accessible to bus patrons via a stairway and/or ramp.
Bus shelter- A structure constructed near a bus stop to provide seating and protection from the weather for the convenience of waiting passengers.
Bus stop- A place where passengers can board or alight from a bus, usually identified by a sign.
Carpool- An arrangement in which people share the use and cost of a privately owned automobile in traveling between pre-arranged destinations.
Community Shuttle- a feeder service to the main transit vehicle service from inner localities to the main roads
Commute service- Express bus service during commute periods.
Commuter- Passenger who travels between two points regularly.
Connection- The linking of multiple transit trips together to reach a final destination. May include more than one mode of travel (bus, ferry, rail, etc.).
Deckhand- Crew member responsible for assisting staff to ensure the safe passage of passengers.
Departure time- The time a bus or ferry leaves from any designated timepoint on a schedule.
Destination sign- A sign on a transit vehicle indicating the route number, direction, or destination of the vehicle. Also known as a head sign.
Detour- A change in a route due to construction, flooding, obstruction, etc., that causes a vehicle or vessel to be re-routed.
Division- Refers to a garage and yard facility where buses are stored, maintained, and dispatched into service.
Fare- The cost for an individual to be a passenger onboard a transit vehicle or vessel.
Farebox- A piece of equipment controlled by a bus driver or ferry terminal assistant to collect the appropriate transit fare.
Fixed route transit- A system in which transit vehicles follow a schedule over one or more prescribed routes.
Fleet- A transit system's vehicles and vessels.
Frequency of service- The number of transit vehicles on a given route, moving in the same direction, which pass a given point within a specified interval of time, usually expressed in minutes. Also known as headway.
Headsign- A sign on a transit vehicle indicating the route number, direction, or destination of the vehicle. Also known as a destination sign.
Headway- The number of transit vehicles on a given route, moving in the same direction, which pass a given point within a specified interval of time, usually expressed in minutes. Also known as frequency of service.
High capacity- A term used to indicate a bus or vessel capable of carrying more customers than other buses or vessels in our fleet.
Layover time- Time built into a schedule between arrival and departure for bus drivers to rest; minimum times are set by union contract. Layovers normally occur at each end of a route to allow for a driver's break and schedule recovery, but they may be scheduled at other points to allow for timed transfer connections.
One-way- Term used to designate the traveling in one direction from an origin to a destination. Two one-way trips can constitute a round trip.
Zone- Area delineated by boundaries within our service area used to determine passenger fare.
What are transit vehicles
Atransit vehicle, which is sometimes also known as public bus, city bus or simple a commuter bus, is a vehicle most often used as a transport vehicle for short distances in a city or between two points in a city. The roles that these transit vehicles play may differ from operator to operator and region to region. The changing technologies in bus building and with rapid urbanization, transit vehicles have undergone several changes including that of inclusion of features that ensure that the vehicle adapts itself to locally prevalent conditions of the region. These features are quite different from other private or public transport buses.
Transit vehicles are designed and built in such a manner so that they provide efficient services to the general public. Transit vehicles could either be used to ferry school kids or be used for the general public in transporting them from one location to another in a city. They are generally fit for operations in the urban or semi-urban areas, locations which have defined bus stops and usually have a short-distance role. This is as opposed to long haul buses which are used for coach transport that ply over long-distances. In addition, transit vehicles have additional features such as designed for the physically challenged, are usually low-floored and so on. Transit vehicles are usually characterized by the presence of all or any of the following i.e. less amount of storage area for luggage, seats that either bucket or bench and do not have head-rests, notice boards for the destination which are either static or electronic and finally, a legally allowed passenger capacity.
Transit vehicles - Operations
Modern transit buses are also increasingly being equipped withpassenger informationsystems, multimedia entertainment/advertising, and passenger comforts such as heating and air-conditioning (as opposed to historically where actually necessary). Some industry members and commentators promote the idea of making the interior of a transit bus as inviting as a private car, recognizing the chief competitor to the transit bus in most markets. As they are used in a public transport role, transit buses can be operated by publicly run transit authorities ormunicipal bus companies, as well as private transport companies on a public contract or fully independent basis. Due to the local authority use, transit buses are often built to a third-party specification put to the manufacturer by the authority. Early examples of such specification include the 'Greater Manchester'Leyland Atlantean, and DMS-class LondonLeyland Fleetline. New transit buses may be purchased each time a route/area is contracted, such as in theLondon Busestendering system.
The operating area of a transit bus may also be defined as a geographic metropolitan area, with the buses used outside of this area being more varied with buses purchased with other factors in mind. Some regional-size operators for capital cost reasons may use transit buses interchangeably on short urban routes as well as longer rural routes, sometimes up to 2 or 3 hours. Often transit bus operators have a selection of 'dual-purpose' fitted buses that is standard transit buses fitted with coach-type seating, for longer-distance routes. Sometimes transit buses may also be used asexpress buseson a limited-stopping or non-stop service at peak times, but over the same distance as the regular route.
Transit vehicle research
The design of collision warning systems recognizes the relative speed, orientation, and distance between the instrumented vehicle and nearby objects, and the influence of environmental factors, such as weather, lighting, and roadway conditions. However, research and development efforts to date have focused on light vehicles and long-haul trucks ignoring the particular needs of transit, such as the requirement to operate in restricted spaces and the presence of pedestrians. Existing side collision warning systems are typically unable to resolve smaller objects that are common in the transit buses' operational domain, such as pedestrians, bicycles, and baby carriages. Furthermore, existing systems require multiple sensors for adequate coverage of the side of a 40ft transit bus.
A collaborative effort involving Carnegie Mellon University, the Port Authority Transit (PAT) of Allegheny County, the Pennsylvania Department of Transportation, and the USDOT
Federal Transit Administration addressed these issues in an earlier study. The goal of this work was to investigate, develop, and test performance specifications for a side collision warning system that can reliably detect these smaller objects using, preferably, a single sensor per side of the bus. A preliminary review of publicly available data and informal interviews of bus drivers suggests that pedestrians present a significant problem. For example, a study in Washington State (Wessels et al.) found that 127 out of 11160 pedestrian collisions in the five-year period 1990-1995 were bus related. While this only represents 1.1% of pedestrian collisions, it is significant, as nationally only 0.26 % of vehicle miles of travel are by bus (Bureau of Transportation Statistics, 1998). Furthermore, an even larger proportion (3.1%) of the bus related pedestrian collisions resulting in fatalities is reported in the Washington State study (Wessels et al.) suggesting that the bus pedestrian accidents may be more severe. The high cost of accidents, the importance of safety in the public perception of transportation alternatives, and recent advances in sensors, automated and intelligent vehicles, vehicle control and driver interfaces mean that there is an opportunity to address this problem. In the paper we review past efforts at safety related transit improvements, and related studies for light vehicles and long-haul trucks. The paper also evaluates experiences on Port Authority buses in Pittsburgh in terms of observing pedestrian incidents and experiences with an existing sensor for side collision warning. Finally the paper summarizes opportunities related to improving side collision warning systems and describes a plan for future research.
Mishaps and accidents
The Washington State study found that bus pedestrian incidents had some common themes (Wessels et al.). The number of bus collisions are highest during morning and evening rush hours and are predominantly on city streets. In addition, most pedestrians are in the roadway when struck; small percentages are on the sidewalk or shoulder. In the most common case, there was no violation by the pedestrian; second most common was inattention; third was alcohol. In another case, there was no violation by the driver; inattention and fail to yield were also significant contributing circumstances. Distribution of crashes by bus driver age peaks in 35-44 and 45-54 decades, which may indicate that inexperience, is not the major factor.
Interviews with some of the drivers revealed something about the nature of crashes involving buses. The theme in these conversations is a major concern with preventing pedestrian accidents.
Although there are relatively few pedestrian accidents, they are the most likely to cause serious injury. They are also among the hardest to prevent. Bus drivers do not trust the common sense of pedestrians - the anecdotal evidence is full of people walking in front of busses, even people walking into the side of a bus, knocking themselves down, and being run over by the rear wheels.
Pedestrians are significantly harder to see than vehicles since they are smaller, move unpredictably, and can be found in areas close to the bus where the driver has limited visibility.
Such areas include the front of the bus in the area blocked from view by the electronic fare box, and the area immediately behind the front passenger door. Beyond pedestrians, there is concern with bicycles (including fast riders going around the bus on the curbside). There is some concern with hitting fixed objects, but the general feeling is that those accidents are most likely due to driver inexperience. There is more need for sensors along the front half of the side of the bus, short-range sensors looking forward, and perhaps sensors watching the ground in front of the rear wheels; and less need for sensors at the rear corners.
To enable us to go beyond the anecdotal evidence, PAT has provided us with their database of liability claims. This database includes comprehensive data for claims and crashes since 1997. For the period January 1997 to May 1999 the database includes over four thousand records. PAT includes in the claim record a field "Nature of Accident." Claims categorized as "Bus & Pedestrian" in this field were extracted. A total of 141 records were extracted representing 90 unique incidents. These incidents are largely clustered in Downtown Pittsburgh and Oakland (an area with universities and hospitals and heavy pedestrian traffic) as shown in Figure 1. Recognizing that the downtown area is less than one square kilometer, there are approximately 19 incidents in a 29-month period.
Our current understanding is that pedestrian strikes are the most worrisome bus related incidents, both because of the serious consequences, and because people move unpredictably and are therefore harder to track. Drivers voice pedestrian related concerns during interviews that fall into one of two categories: Interactions with the bus when the bus is just starting to move after having stopped. When the bus is moving at speed, the driver can see pedestrians as he or she approaches, and they are unlikely to move fast enough to get into the bus blind spots before the bus clears the area. When the bus is stopped, pedestrians can more easily move into blind spots, and are less likely to be concerned about the bus moving. Moving objects (people) when the bus is stationary. This is significant, because moving object detection from a static platform is a much easier problem than generic person detection. To better understand these issues we identified what the driver sees. Based on these results, the literature review and past experiences we then identified opportunities for address this problem.
Buses have quite limited visibility along the right side, and areas of limited visibility in front where the driver's view is obscured by mirrors, the farebox and a high dashboard. Unfortunately, these areas of limited visibility correspond to the locations where pedestrians are likely to be. Using approximate techniques based on sight lines, areas of limited visibility around a bus were identified. Figure 2 is a schematic showing the “blind spots around the bus”. The left side of the figure is a plan view showing areas obscured by the mirrors and farebox and the areas behind the driver not covered by the mirrors. A person standing or moving in any of the areas defined by straight lines and arcs will not be seen by the driver unless the driver moves his or her head, or the person moves part of their body out of that zone. In reality, this is a three dimensional problem and the right side of the figure presents a longitudinal view showing areas obscured by the mirrors, farebox and dash of the bus. For example, a person of typical height is often partially visible. However, the farebox or dashboard of the bus can easily obscure a short person. This again reinforces our view that concern for pedestrian safety is paramount, and is a much different problem than lane-change-merge countermeasures.
This emphasis on pedestrian accidents corresponds with our initial intuition, but is somewhat different from the emphasis suggested in other studies and discussion. We believe that the operating environment of a transit bus is very different from the environment of a light vehicle or long-haul trucks and that the emphasis of this project should be quite different than the emphasis of the NHTSA lane-change/merge countermeasures program. However, there are also many opportunities to build on past and on-going work.
The literature on pedestrian detection, lane-change / merge collision detection, and human factors includes applications to vehicle-vehicle collisions, pedestrian detection, object detection and avoidance, and specific applications related to school buses. This section presents a brief synopsis of the state of the art organized in terms of an overview, sensors and detection, interfaces, and commercially available collision warning systems (CWS).
Several reports and papers address experience with existing systems or system components.
Other reports present results from testing commercial and prototype systems. For example, eleven CWS for lane change, merge and backing were evaluated using operational and performance data. The data were obtained on a controlled test track and on public roads. Performance capabilities varied widely (Taknedge et al., 1995).
The foundation of much of the relevant research related to vehicle-vehicle side collisions is the functional (high level) goals (for both vehicles and infrastructure) for eliminating lane change, merge and backing crashes (Young et al., 1995). To understand side collisions, they develop a taxonomy for vehicle-vehicle crashes as follows:
leaving a parking place
both changing lanes
They then derived the following functional goals:
1. To alert the driver to the presence of vehicles in adjacent lanes prior to initiation of lane change maneuvers
2. To alert the driver of drifting vehicle motion.
3. To alert the driver to the presence of rapidly approaching vehicle in adjacent lanes.
4. To alert the drive to the presence and movement of vehicle two lanes over.
1. To heighten the awareness of drivers as they approach a merge
2. To provide situational awareness during merging
These functional goals have been translated into preliminary performance specifications (TRW, 1995). These specifications build on the concept of a CWS as shown in Figure 3, and address the goals identified above as shown in Table 1. The concepts used are relevant to this problem since buses also change lanes and merge, but need to be expanded to reflect the importance of bus pedestrian crashes.
Sensors and Detection
Pedestrian detection has been the focus of several studies as follows:
The City of Portland tried three different kinds of sensors for monitoring pedestrians in crosswalks. They had limited success with sonar, mostly due to their mounting angles. They
preferred infrared for narrow-beam short-range detection and microwave Doppler radar for longer range detections, and engineered a system that gave nearly 100% coverage (Beckwith and Hunter-Zawarski, 1997).
The Society of Automotive Engineers (SAE), with sponsorship from NHTSA, used Doppler radar to detect children around school buses. Because the technology detects movement, it only works with a stopped bus and a moving child. Two different systems were tested; both included forward and sideways looking components. The results are regarded as generally encouraging (Johnston et al, 1996).
Sonar was used to detect walking people by looking for rhythmic patterns of the frequency expected in typical gaits (Sabatini and Colla, 1998).
The Nissan Advanced Safety Vehicle includes an infrared camera to detect pedestrians in front of the vehicle. If a hot spot is detected, one of several LEDs on the dash lights up, drawing the driver's attention to the right direction (Sugasawa et al., 1996).
A detailed description of how thermal sensors work is presented (Everett Infrared) in a review of the physics of sensors. The most interesting discussion is that most sensors are better at detecting changes in temperature, rather than absolute temperature, so detecting moving objects is easier than detecting stationary bodies. A NHTSA study focuses on radar and lidar as technologies, as well as determining the state of the art in digital signal processing, for lane change, merging and backing. This study uses the preliminary performance specifications from other tasks of the NHTSA study as a benchmark (Moffa et al., 1996). Carnegie Mellon researchers are developing a stereo based pedestrian detection device that uses neural networks to process the data (Zhao and Thorpe, 1999). Cadillac is using an infrared imaging system for night vision enhancement. The system identifies pedestrians and displays an enhanced image on a heads-up display (http://archive.abcnews.go.com/sections/tech/CarTech/cartech980902.html)
The literature on interfaces includes discussion of concepts and configurations, and presentation of results from experiments to evaluate different devices. For example, Campbell et al. (Campbell et al., 1996) identified and evaluated driver vehicle interface designs for side object detection systems using static mock-ups and displays in a driving simulator. Variables included format, location, and symbols. The study concluded that three types of information are valuable status indication at vehicle start-up, caution alert under “no intent to turn' situations, and hazard alert under “intent to turn” situations (including directional information)
Some basic common sense points are worth noting as follows:
• color of indicator alone is not sufficient, because there may be color blind drivers;
• brightness of indicators needs to vary from high (in direct sunlight) to more subdued (at night); and
• hold time and response time are critical parameters, and depend both on the sensor and on driver behavior. (Hyland, 1995)
Similarly experiments for rear end collisions were used to develop some general principles including the value of the concept of a two-stage alert; alternative sizes, shapes, and colors of icons; and sound specifications (Landau, 1996). More specifically, a NHTSA study puts different visible and audible interfaces on two different vehicles, and drove them in a variety of traffic conditions. Specific recommendations for how much the driver should be able to adjust loudness, how bright LEDs should be, etc. are included in the paper (Mazzae et al., 1995). In another study, driver interfaces for side looking collision-warning systems, rear looking collision-warning systems and systems that enhance the driver's ability to see object in the rear of the vehicle were assessed. The assessment was used to develop a preliminary set of driver interface performance specifications (Mazzae and Garrott, 1996).
Of a more general nature is a summary of human factors research to date and human factors research needs related to lane-change and merge collision avoidance, including obstacle and pedestrian detection presented in (FHWA, 1998).
Commercially Available Side Collision Warning Systems.
A summary of vendor-supplied information about their most current warning systems as of November 1998 is provided in (AssistWare Technologies, 1998). The data was collected specifically for application to transit buses. ITS America data and Internet searches were used to identify vendors who were then asked to respond to a questionnaire that documented:
• the operation of the system,
• the applicability of the system to transit,
• the sensing technology used,
• previous experience with using the technology for side collision warning, and
• lead-time and warranty information.