Method and apparatus for trajectory shape generation for autonomous driving vehicles

Abstract

An apparatus for controlling the direction and speed of an autonomous vehicle or driver-assisted autonomous vehicle (AV). A GPS and map module receives a starting position and a destination position of the AV. A plurality of sensors identify a current lane and a proposed lane of the AV. A database of AV baseline maneuver profiles for controlling one or more of the direction and speed of the AV is provided. A trajectory profile generator module generates a planned path of the AV having a lateral acceleration of less than 2 Hz based on the starting position, the destination position, the current lane and the proposed lane of the AV and the AV baseline maneuver profile selected from the database. A steering control module controls the direction of the AV based on the generated AV planned path, and a supervisory control module controls the speed of the AV based on the generated AV planned path and an inner ear constraint.

Description

[0001] Priority claim

[0002] This patent application is related to and claims priority from Provisional Patent Application No. 62 / 859,649, filed on June 10, 2019, entitled “Optimal Trajectory Shape Generation for the Minimization of Motion Sickness in Autonomous Vehicles,” the contents of which are incorporated herein by reference. Technical Field

[0003] Embodiments of the present invention relate to autonomous vehicles, and more particularly, to controlling the trajectory of an autonomous vehicle to minimize motion sickness of passengers in the autonomous vehicle. Background Art

[0004] An autonomous vehicle (also called an autonomous vehicle (AV), connected autonomous vehicle (CAV), driverless car, or robo-car) is a vehicle that can sense its environment and move safely with little or no human input. An autonomous vehicle uses a combination of sensors such as video, radar, lidar, sonar, GPS, odometry, and inertial measurement units to sense its surroundings. The control system interprets the sensory information to identify appropriate navigation paths as well as obstacles and relevant landmarks.

[0005] Autonomous vehicles can make their passengers motion sick. Motion sickness is related to the frequency, level, and direction of external accelerations imposed on the passengers, and to the passengers' lack of anticipation of maneuvers performed by the autonomous vehicle. Part of the passengers' lack of anticipation is based on how the autonomous vehicle is maneuvering, i.e., the shape of the velocity vs. time graph of the autonomous vehicle's acceleration and the attitude (i.e., three-dimensional orientation) of the autonomous vehicle. Both the time-optimal paths and constant curvature turns used in autonomous vehicles are not the way people maneuver vehicles. What is needed is a "passenger comfort" optimal path, where reduced potential for motion sickness is the property to be optimized. BRIEF DESCRIPTION OF THE DRAWINGS

[0006] The embodiments are illustrated by way of example and not limitation and will be more fully understood upon reference to the following detailed description considered in conjunction with the accompanying drawings, in which:

[0007] Figure 1 is a functional block diagram of an embodiment of the present invention; and

[0008] Figure 2 is a description of the geometric constraints used when generating trajectories. DETAILED DESCRIPTION

[0009] Embodiments of the present invention control autonomous vehicle trajectory and speed to reduce the likelihood of motion sickness for passengers. In particular, embodiments of the present invention control a steering trajectory between dynamic states of position (direction), velocity, and acceleration of an autonomous vehicle. The dynamic state of the autonomous vehicle in this case consists of changes in the vehicle's embedded longitudinal axis (from rear to front) (controlled by propulsion and braking) and changes in the vehicle's lateral or cross-axis (controlled by steering). Multiple interdependent controls for vehicle speed and direction are used to create the desired motion sickness reduction trajectory. Controls include, but are not limited to, controlling the jerk (i.e., changes in acceleration) component of the dynamic state (including lateral and forward motion) within the path constraints of the desired direction of travel.

[0010] Speed ​​and direction control are managed by a supervisory control module that constrains the time-dependent lateral acceleration and jerk to within a frequency range that could disturb the inner ears of passengers in an autonomous vehicle. Figure 2 , examples of autonomous vehicle maneuvers to be controlled in this context include actions such as turning from one street to another or changing from one lane 220 to another (such as changing lanes or turning from one street to another). Lane 220 is defined herein as the width of an area of ​​a road or surface on which an autonomous vehicle can be driven. More broadly, a maneuver is any change in course or speed, and a trajectory is a path taken between one or both of these states. The embodiments are primarily directed to autonomous vehicles, but may also be applicable to driver-assisted autonomous vehicles.

[0011] Reference Figure 1 , an embodiment of the present invention 100 receives input from three sources: a GPS and map module 101, which provides the autonomous vehicle with a current or selected (or chosen) starting location and a selected destination location with reference to a digitized map; a baseline maneuver profile database 103 for the autonomous vehicle, which provides basic information for controlling the direction, speed, and acceleration of the autonomous vehicle, such as for changing lanes, turning left or right, or entering or exiting a highway; and sensor input 104 from sensors, such as forward-looking sensors and optional side-looking sensors, which identify the autonomous vehicle's current lane and turning environment and the autonomous vehicle's proposed lane (if different from the current lane).

[0012] The inputs to the GPS and map module 101 (e.g., provided by or for a passenger) include the starting or current location of the autonomous vehicle (location A) and the destination location (location B). The GPS and map module 101 identifies a plan for getting the autonomous vehicle from location A to location B and provides it to the trajectory profile generator module 105. As the autonomous vehicle travels, the sensors 104 (e.g., one or more monocular or binocular cameras, radars, lidars, GPS transceivers, or a combination of sensors) sense approaching turns or identify possible, proposed, or planned paths (based on inputs received from the GPS and map module 101) or lanes 220, and capture turn and lane 220 parameters (including road conditions, any vehicle congestion or other types of road congestion (pedestrians, cyclists, road construction, obstacles, signs), and upcoming turns to be negotiated based on the GPS and map inputs) and pass this information to the trajectory profile generator module 105. The motorway lane requirements from the baseline maneuver profile database 103 may also be used by the trajectory profile generator module 105. The trajectory profile generator module 105 selects a maneuver profile from the baseline maneuver profile database 103 and calculates a path with low lateral acceleration (e.g., less than 2 Hz) to complete the maneuver. (Keeping the primary acceleration forces in a vertical direction with reference to the passenger's anatomy rather than in a lateral direction is accomplished by using coordinated turns such as in a two-wheeled vehicle or in an autonomous vehicle (the coordinated turns rotate the passenger to keep the sum of gravity and centrifugal forces aligned so that the passenger experiences only a vertical sensation)).

[0013] The calculated path with low lateral acceleration is sent to the steering control module 107 and the supervisory control module 106. The supervisory control module 106 uses the geometry of the path, the inner ear constraints 102, the speed limit obtained from the GPS and map module 101, and environmental information (e.g., road conditions, traffic congestion) to select the speed of the autonomous vehicle. The speed is transmitted to the trajectory profile generator module 105 to set the lane tolerance 205 in subsequent iterations of generating trajectories for the autonomous vehicle, and to the speed control module 108. The inner ear constraints 102 consist of the frequency levels in the inner ear cross spectrum and the decay time of the interference history.

[0014] According to one embodiment, the Apriori trajectory generator module uses a set of tables to place points between lane boundaries for the total turn angle range. These data are generated based on the autonomous vehicle design during the development phase of the system design and include the percentage of progress through the turn when the number of defined points changes from 3 to 2.

[0015] According to embodiment 200, Figure 2The direction of travel or turn direction (DOT) 225 and lane 220 parameters depicted in the plan profile generator module 105 include the lane 220 and the trajectory of the center of the planning area 210 or the desired path (if a planning area exists). The planning area 210 is defined herein as the width of the lane 220 minus the width of the lane tolerance 205 (one or both of the left and right lane tolerances, if they exist) and minus the width of the autonomous vehicle ("vehicle width 215"). The lane tolerance 205 is defined herein as the safety barrier between the vehicle's furthest maneuver and the edge of the lane 220. The lane tolerance 205 changes with vehicle speed and lane 220. The planning area 210 is a path range that the trajectory profile generator module 105 can use to design a path so that the autonomous vehicle travels with minimal inner ear disturbance, or so that the autonomous vehicle travels with inner ear disturbance that at least remains below a threshold that is considered to cause motion sickness. Inner ear disturbance changes with path design and vehicle speed. When driving an autonomous vehicle on a narrow road, the planning area 210 may shrink to zero and the only parameter that can be used to control lateral acceleration is the autonomous vehicle speed.

[0016] Note that if the width of lane 220 decreases in width to less than twice the width of lane tolerance 205 and autonomous vehicle width 215, planning region 210 may be reduced to a line. Additionally, if the width of lane 220 decreases to less than vehicle width 215, planning region 210 may reach an end point. Vehicle width 215 is defined herein as a cross section of the vehicle orthogonal to its velocity vector relative to lane 220 and is a parameter of vehicle configuration and attitude. Vehicle width 215 determines a limit distance that the center of a generated trajectory may approach the inside edge of lane tolerance 205. A generated trajectory is defined herein as a path that satisfies the geometric planning constraints and reduces or minimizes the probability of motion sickness.

[0017] The vehicle width 215 of a two-wheeled autonomous vehicle may vary with attitude (unless it is spherical, i.e. a leaning two-wheeled vehicle is wider than an upright two-wheeled vehicle). The lane 220, along with the vehicle width 215 and lane tolerance 205 requirements, are used to calculate the planning region 210. The lane tolerance 205 is defined herein as the portion of the lane 220 at the edge of the lane 220. The lane 220 minus the lane tolerance 205 (the lane tolerance on one or both sides of the lane 220) defines the planning region 210. The planning region 210 provides a free range to change the trajectory of the autonomous vehicle from the center of the lane 220.

[0018] The lane tolerance 205 requirement is the distance that the vehicle must maintain from the edge or boundary of the lane 220 (left edge or right edge, single lane boundary or double lane boundaries). According to an embodiment, the lane tolerance 205 requirement varies with the width of the lane 220 width and the speed of the vehicle. A wider lane 220 incurs or allows for a larger lane tolerance 205 requirement. The higher the speed of the autonomous vehicle, the greater the lane tolerance 205 requirement because at higher speeds higher accelerations occur and corrective action requires a larger tolerance.

[0019] According to an embodiment of the present invention, there are two basic types of maneuvers: changing destination, including changing lanes, choosing directions at intersections, turning from one highway to another, and exiting or entering a highway; and following a curved road. Roads with occasional turns, bends, or road changes can be addressed more aggressively because motion sickness is a cumulative condition. The history of introduced acceleration disappears over time. However, driving on a mountain road with many bends may require a reduction in speed to prolong the acceleration event and also reduce the magnitude of the lateral acceleration (i.e., the maximum lateral acceleration of the autonomous vehicle).

[0020] These factors require a two-level path planning approach as follows. Each lateral acceleration event that results in low-frequency lateral acceleration (below 2 Hz) and has a magnitude of lateral acceleration above human sensitivity is recorded along with its severity. The time-weighted sum of the severity is used to set the lateral acceleration level for the upcoming maneuver or planned path.

[0021] The planned trajectory is designed to minimize and smooth its curvature as the autonomous vehicle travels along the trajectory. One embodiment generates a curve with C3 characteristics or higher between the current position and orientation of the autonomous vehicle and the desired position and orientation of the autonomous vehicle. That is, the first derivative of the curvature is at least a continuous function. Further, the second derivative of the curvature may also be limited in absolute magnitude. The continuity of the first derivative of the path curvature (C3 position curve) reduces the frequency components of the acceleration felt by the passenger and further reduces motion sickness. Smoothing of lateral acceleration is achieved by using at least a C3 (continuous third derivative) curve for the path.

[0022] An example of a method to generate a curve with two continuous derivatives (C2) is to use the following third-order spline curve:

[0023] P=A+Bt+Ct 2 +Dt 3 .

[0024] However, the path is generated by using two curves, one in x and one in y. Plotting the curve in the xy lane provides a path for the turn. The first derivative of the composite curve generates the direction, and the first derivative of the direction together with the wheelbase of the vehicle gives the curvature. Passengers in an autonomous vehicle require smooth curvature, which means that at least the first derivative of the curvature is smooth. This requires a path with at least three continuous derivatives, or a C3 curve. This can be generated by using the following fourth-order spline curve:

[0025] P=A+Bt+Ct 2 +Dt 3 +Et 4 ,

[0026] Or by, for example, a curve like this:

[0027] P x =A x +B x t+C x t 2 +D x t 3 +E x t 4 +…

[0028] and

[0029] P y =A y +B y t+C y t 2 +D y t 3 +E y t 4 +….

[0030] It should be further understood that embodiments of the present invention may use C3 or higher curves (including curves generated by versine, etc.) instead of spline curves.

[0031] According to an embodiment, a straight 4th order spline curve may be used, but the process of starting with a 3-point 4th order spline curve and generating a 4-point 4th order spline curve using slightly modified slopes of two parametric curves reduces the overall curvature and lateral acceleration experienced by the passenger.

[0032] The two curves generated are:

[0033] P x (t) and P y (t),

[0034] The parameter t is the distance along the path. The calculation of the coefficients is left to those skilled in the art. The direction and curvature are given by x(t) and P y (t) It is concluded that:

[0035]

[0036] and

[0037]

[0038] The magnitude of the curvature is used to calculate a bound on the velocity to keep the lateral acceleration below the maximum acceleration determined from the motion sickness sensitivity.

[0039] According to an embodiment of the invention, the change in speed is controlled in a similar manner as described above with respect to the change in curvature of an autonomous vehicle. An embodiment further generates a continuous jerk (first derivative of lateral acceleration) function.

[0040] Motion sickness is further reduced if the autonomous vehicle has only two wheels in a row. People are less susceptible to vertical acceleration than lateral acceleration. Two-wheeled vehicles have coordinated turns to maintain balance. This reduces or eliminates lateral acceleration to human sensory receptors, allowing a two-wheeled vehicle to travel faster along a desired path with the same level of motion sickness sensitivity. It was noted above that the width of lane 220, as well as the vehicle width 215 and lane tolerance 205 requirements, are used to calculate the planning area 210, i.e., the free range to change trajectory from the center of lane 220. When embodiments of the present invention are applied to a two-wheeled vehicle, the vehicle width 215 is less than a four-wheeled vehicle, so the planning area 210 is larger.

[0041] If the autonomous vehicle is used as a robotaxi, the motion sickness level should be set at the 10 percentile level (a certain percentage of passengers will experience motion sickness in a given environment), and if it is used as a personal autonomous vehicle, the level can be set at the percentile level of passengers using the vehicle, for example, by turning a knob or touchscreen or otherwise selecting a driving mode between Comfort, Sport, or Track. The higher the passenger's resistance to motion sickness, the faster the passenger can traverse a path to their desired destination.

Claims

1. A device for controlling the driving direction and speed of an autonomous vehicle, comprising: a GPS and mapping module for receiving a starting location and a destination location of the autonomous vehicle and a speed limit, and generating a plan for moving the vehicle from the starting location to the destination location; a trajectory profile generator module for receiving the plan and calculating, as the autonomous vehicle is traveling, a path including a maneuver for moving the vehicle from the starting location to the destination location, wherein calculating the path as the autonomous vehicle is traveling includes: tracking each lateral acceleration event having a magnitude above a human sensitivity level and a corresponding severity; calculating a time-weighted severity sum based on corresponding severities of the tracked lateral acceleration events; and setting a lateral acceleration level for the path or upcoming maneuver based on the time-weighted severity sum; A supervisory control module (106) is used to: receiving the calculated path and selecting a speed of the autonomous vehicle based on the geometry of the calculated path, inner ear constraints including frequency levels in a human inner ear cross-spectrum and decay times of inner ear interference history, the speed limit, and environmental information; and The calculated path and the selected speed are received, and a lateral acceleration of the vehicle is reduced based on the calculated path and the selected speed.

2. The device according to claim 1, wherein: The trajectory profile generator module for receiving the plan and calculating the path of the maneuver for moving the vehicle from the starting position to the destination position as the autonomous vehicle travels includes the trajectory profile generator module for generating a curve having C3 characteristics or higher characteristics.

Images