Kenji Tenmoku

A fast algorithm for finding better routes by AI search techniques

The shortest path problem is one of the most fundamental problems applicable in various fields, and has close relation to route navigation systems. This paper surveys algorithms for the two-terminal shortest path problem and proposes bidirectional A* algorithm based on a new approach. This algorithm is suitable for finding not only the shortest route but also better routes. The efficiency and the property of these algorithms are discussed through experiments applying them to an actual road network.<<ETX>>

Cooperative steering system based on vehicle sideslip angle estimation from side acceleration data at percussion centers

This paper considers how a manual steering scheme can be combined with automatic steering for a cooperative steering system which yields better stability, together with path following capability. A new version of the combination is proposed based on a course error differential equation at a percussion center with respect to rear-wheels. The automatic steering includes the feedback of vehicle sideslip angle which is difficult to measure by an ordinary device. An adaptive estimation scheme is proposed to feedback the sideslip angle. Driving simulator experiments demonstrate good estimates and good performance of the cooperative steering system.

A cognitive animation simulator for a group of car drivers

Group behavior of car drivers is considered. A cognitive simulator to animate car movements is constructed. Driving situations are first analyzed into basic ones by a tree model containing case and sequence structures. Driving activity trees to be linked to basic situations are derived from a general activity tree. The resultant tree is traced at each sampling time. The tracing is controlled by a rule base. Rule conditions and animation results are demonstrated.<<ETX>>

Accuracy of travel time necessary for DRGS and evaluation of measuring algorithm

The dynamic route guidance system (DRGS) using real-time, dynamic information has been identified as a necessity for alleviating todays chronic traffic congestion. Current traffic information, and accurately determining and predicting travel time data in particular, are important factors directly affecting DRGS viability. The effect of the estimated (current) link travel time precision on DRGS performance, and the conditions for practical DRGS implementation, were evaluated. It was determined that to effectively provide optimum route guidance and travel time information, the estimation precision of the link travel time must be within 10% of the actual travel time, and the prediction precision must be a fair estimation with a maximum 25% margin of error for destinations 90 minutes ahead. An intelligent travel time prediction method executed by the on-board system was studied and evaluated. This method was shown to accurately measure the travel time for approximately 90% of all links during normal travel, thus virtually eliminating abnormal travel and providing sufficient practical performance.

Sliding and adaptive sliding mode controllers for automated, longitudinal car driving

Longitudinal controllers are derived for an automated car driving. Two types of information sources are considered: the first-order information includes location and speed of following and leading cars, while the second-order information contains acceleration/deceleration in addition to the first-order information. A proper headway distance between the two cars is defined appropriately, sliding mode and adaptive sliding mode control theories become applicable, and longitudinal controllers are derived. The resultant controllers are proven to be stable and robust. The acceleration/deceleration information available through the modern inter-vehicle communication and sensing technologies significantly improve the control performance. The adaptive sliding mode controller yields better results because it includes a stable estimation mechanism of the maximum disturbance value.

Rule-based cognitive animation simulator of fallible drivers around intersection

Group behaviors of car drivers around a double-lane intersection are animated by a cognitive simulator. A situation tree is constructed to analyze the driving situations into successively more basic ones. Driving activities specific to each basic situation is analyzed as a subtree of a general activity tree. A resultant expanded situation tree is traced at each sampling time. If-then rules are used to represent feasibility conditions of each microscopic transition from a parent node to a child node in the expanded tree. Each trace path is a sequential arrangement of these transitions. If-part conditions are clarified and animation results are demonstrated. Traffic accidents are initiated by mistakes (if-part errors) and slips (then-part errors). Mistakes are omission and commission errors during intention formations. The commission errors are further classified into external and internal errors. Internal errors are risk-prone because they are related to parameters in if-part conditions attributable to drivers personality.<<ETX>>