202
Sensemaking in Safety Critical and Complex Situations
Vehicles ( DMV) in 2019 shows that other 65 companies currently testing level 4
technology still have frequent rear-end collisions at signalised junctions. They also
have trouble ( and reported accidents) entering a motorway from the ramp. AVs have
not yet learned the “ nudging” that ordinary drivers do to see if traffic on the motorway yield and let you in.
Experience from the autonomous shuttle buses: For the pilots, it was mandatory to report incidents and accidents. No persons were injured, and only minor
technical issues and malfunctions were reported. The following issues were revealed:
– Snow, heavy rainfall and fog are challenging for the sensors.
– Vegetation and light poles along the route of the bus is challenging as they
interfere and disturb the sensors at times.
– The buses run along the same “ track” with narrow wheels, causing significant wear and tear on the road along this track.
– Cyclists passing near the bus makes the bus stop abruptly.
These issues are related to the predefined operational envelope surrounding the vehicle, leading to abrupt stops when violated. As pointed out by Jenssen et al. ( 2019),
AVs lack a sense of self, and software and sensors are still not designed to account
for the discrepancy in the same way human drivers are able to.
When applying for testing, a mandatory risk assessment was carried out. The main
risks listed were related to passenger injury as a result of an abrupt stop where passengers inside the bus are unprepared and can be harmed by falling. R isk-reducing
measures are lowering the speed, installing seat belts, limiting the number of passengers and adding road signs.
AGVs at St Olav: A total of 1 00–130 minor incidents per year have been
reported. Yearly, each AGV experiences around 15 emergency stops ( Johnsen et al.
2019), where components must be changed. Reported incidents are minor crashes as
a consequence of faulty navigation due to objects placed in the route, summarised in
Johnsen et al. ( 2019). From interviews with the operators of the AGVs, the following
main issues are identified:
– The AGVs ability to adapt to the surrounding infrastructure
– Keep the track of the AGVs clear of objects
– Make objects visible to the AGV: the AGVs are not able to detect all obstacles due to the sensor range
– Establish a control room with proper HMI design
– Maintain the interface to cyber physical systems: software updates has led
to problems ( due to poor testing and multiple vendors.)
Lessons Learned
Vehicle automation can enhance safety but also introduces new risks due to poor
technical implementation and the need for rapid response from the human actor. This
is especially the case with SAE automation levels 2 and 3.
The accident data collected so far with automation ( AGVs and level 1–4 vehicles)
indicate safety hazards of human factors and technical issues, i.e. obstacle detection
Sensemaking in Safety Critical and Complex Situations
Vehicles ( DMV) in 2019 shows that other 65 companies currently testing level 4
technology still have frequent rear-end collisions at signalised junctions. They also
have trouble ( and reported accidents) entering a motorway from the ramp. AVs have
not yet learned the “ nudging” that ordinary drivers do to see if traffic on the motorway yield and let you in.
Experience from the autonomous shuttle buses: For the pilots, it was mandatory to report incidents and accidents. No persons were injured, and only minor
technical issues and malfunctions were reported. The following issues were revealed:
– Snow, heavy rainfall and fog are challenging for the sensors.
– Vegetation and light poles along the route of the bus is challenging as they
interfere and disturb the sensors at times.
– The buses run along the same “ track” with narrow wheels, causing significant wear and tear on the road along this track.
– Cyclists passing near the bus makes the bus stop abruptly.
These issues are related to the predefined operational envelope surrounding the vehicle, leading to abrupt stops when violated. As pointed out by Jenssen et al. ( 2019),
AVs lack a sense of self, and software and sensors are still not designed to account
for the discrepancy in the same way human drivers are able to.
When applying for testing, a mandatory risk assessment was carried out. The main
risks listed were related to passenger injury as a result of an abrupt stop where passengers inside the bus are unprepared and can be harmed by falling. R isk-reducing
measures are lowering the speed, installing seat belts, limiting the number of passengers and adding road signs.
AGVs at St Olav: A total of 1 00–130 minor incidents per year have been
reported. Yearly, each AGV experiences around 15 emergency stops ( Johnsen et al.
2019), where components must be changed. Reported incidents are minor crashes as
a consequence of faulty navigation due to objects placed in the route, summarised in
Johnsen et al. ( 2019). From interviews with the operators of the AGVs, the following
main issues are identified:
– The AGVs ability to adapt to the surrounding infrastructure
– Keep the track of the AGVs clear of objects
– Make objects visible to the AGV: the AGVs are not able to detect all obstacles due to the sensor range
– Establish a control room with proper HMI design
– Maintain the interface to cyber physical systems: software updates has led
to problems ( due to poor testing and multiple vendors.)
Lessons Learned
Vehicle automation can enhance safety but also introduces new risks due to poor
technical implementation and the need for rapid response from the human actor. This
is especially the case with SAE automation levels 2 and 3.
The accident data collected so far with automation ( AGVs and level 1–4 vehicles)
indicate safety hazards of human factors and technical issues, i.e. obstacle detection