Designing Multimodal Assistance Systems for Child Cyclists

Learning to ride a bicycle is an essential element of children's development in many western countries. This is a time-consuming process, which often requires regular training to obtain necessary motor and perceptual-motor skills. Since children are physically and mentally developing, their abilities to sufficiently cope with traffic situations are limited, which often leads to low situational awareness and road accidents. For example, it might be challenging and mentally demanding for a 9-year old child to cycle on a busy road in a city center, have a control over a bicycle and pay attention to road hazards.

Our approach for supporting child cyclists lies on a multimodal augmentation of bicycle and helmets with multimodal cues to increase safety without adding additional cognitive load and distraction. Based on the Multiple Resource theory, we investigate the usage of visual, vibrotactile, and auditory modalities to design assisting systems. We explore a combination of these modalities on various locations on a bicycle and helmets. These cues represent warnings, navigation, lane keeping assistance, and traffic behavior recommendations. For the visual modality, we studied the use of ambient light, augmented reality indicators, and icon-based projection in a helmet, ambient light on a bicycle, and on-road laser projection. We explored vibrotactile feedback on a handlebar's grips and a saddle. Finally, we investigated the auditory feedback on a bicycle and in helmets as speech- and beep-based instructions. We conducted the experiments in two conditions: lab experiments in an indoor bicycle simulator and controlled test-track experiments on an outdoor practice test track. We also used Human-centered Design as the method to involve users in the iterative evaluations. Additionally, we examined children's cycling performance and perception of assisting cues in the presence of visual and auditory distraction tasks.

The results of this work show that multimodal assistance systems are suitable to support child cyclists without mental overload and distraction. We found that children spent significantly more time perceiving visual than auditory or vibrotactile cues. When priming stop actions, reaction time was shorter when all three modalities were used simultaneously. Additionally, we found that auditory navigational cues were the most understandable and the least prone to navigation errors. However, light and vibrotactile cues might be useful for educating younger child cyclists. In particular, we found the combination of ambient light in the helmet and vibration on the handlebar can be the most suitable solution for represent lane keeping cues on the roads with missing infrastructure. Additionally, we have shown that a icon-based cues in HUD were the most effective to remind children about the safety gestures. Based on the results of this work, we envision the combination of four signal types, i.e., warnings, navigation, lane keeping cues and safety gesture reminders, in the general assistance system for child cyclists.