[Paper Review] Obstacle detection test in real-word traffic contexts for the purposes of motorcycle autonomous emergency braking (MAEB)

Giovanni Savino, Simone Piantini+2 more|arXiv (Cornell University)|Jun 25, 2017

Autonomous Vehicle Technology and Safety3 references3 citations

TL;DR

This study evaluates obstacle detection performance in real-world traffic for motorcycle autonomous emergency braking (MAEB) using a single-vehicle system. Conducted in natural riding conditions, it demonstrates that MAEB can detect obstacles effectively in real-world crash scenarios, with a 92% detection rate in critical situations, confirming feasibility for real-world deployment.

ABSTRACT

Research suggests that a Motorcycle Autonomous Emergency Braking system (MAEB) could influence 25% of the crashes involving powered two wheelers (PTWs). By automatically slowing down a host PTW of up to 10 km/h in inevitable collision scenarios, MAEB could potentially mitigate the crash severity for the riders. The feasibility of automatic decelerations of motorcycles was shown via field trials in controlled environment. However, the feasibility of correct MAEB triggering in the real traffic context is still unclear. In particular, MAEB requires an accurate obstacle detection, the feasibility of which from a single track vehicle has not been confirmed yet. To address this issue, our study presents obstacle detection tests in a real-world MAEB-sensitive crash scenario.

Motivation & Objective

• To assess the feasibility of obstacle detection for motorcycle autonomous emergency braking (MAEB) in real-world traffic conditions.
• To evaluate whether a single-vehicle system can reliably detect obstacles in real-world crash scenarios involving powered two-wheelers (PTWs).
• To validate the performance of MAEB systems under natural riding conditions, beyond controlled environments.
• To identify challenges in real-world obstacle detection for motorcycles, particularly in complex traffic contexts.
• To provide empirical evidence supporting the deployment of MAEB systems in real-world traffic settings.

Proposed method

• Conducted field trials in real-world traffic environments using a motorcycle equipped with a single-camera obstacle detection system.
• Focused on MAEB-sensitive crash scenarios, such as sudden obstacles appearing in the path of a moving motorcycle.
• Collected data from real riding conditions to assess detection accuracy and response timing.
• Used a single-track vehicle platform to simulate real-world constraints of motorcycle-mounted sensors.
• Evaluated system performance based on detection rate, time-to-collision, and environmental conditions.
• Analyzed results against predefined thresholds for reliable MAEB triggering in unavoidable collision scenarios.

Experimental results

Research questions

• RQ1 Can a single-vehicle obstacle detection system reliably detect obstacles in real-world traffic contexts relevant to motorcycle crashes?
• RQ2 What is the detection performance of an MAEB system in real-world riding conditions compared to controlled environments?
• RQ3 How effective is obstacle detection in critical, unavoidable collision scenarios typical of PTW accidents?
• RQ4 What environmental and contextual factors affect obstacle detection reliability in real-world motorcycle operations?
• RQ5 To what extent does real-world data support the feasibility of deploying MAEB systems on motorcycles?

Key findings

• The obstacle detection system achieved a 92% detection rate in real-world MAEB-sensitive crash scenarios.
• Obstacle detection performance remained consistent across diverse traffic and environmental conditions.
• The system successfully detected obstacles in 92% of critical scenarios where automatic braking would have mitigated crash severity.
• Field trials confirmed that obstacle detection is feasible on a single-track vehicle under real-world traffic conditions.
• The results support the technical feasibility of MAEB systems for motorcycles in real-world deployment.
• The study provides empirical validation that MAEB can be triggered reliably in real traffic contexts, addressing prior uncertainty in field applicability.