The presence of humans raises new issues in motion and manipulation planning; clearly the presence of humans must be taken explicitly into account in all levels of robots design in order to ensure:
A safe motion, i.e. a motion that in any case would not physically harm/injure the humans in the environment.
A reliable and effective motion, i.e, that achieves the task adequately considering the motion capacities of the robot.
A user friendly motion, i.e, that takes into account a motion model of the human as well as his preferences and needs.
These aspects can be reflected to various levels of robot's design. In the physical design process, making the robot's structure softer using light and soft material will minimize the consequences of a possible impact and making the robot look less threatening will ease its interaction with humans. We claim that the planning is a very important stage to include the human interaction issues. It is obvious that, as the motions and the posture of the robot are calculated by the planner, a motion that takes into account the presence of humans will result a friendly interaction. Our research is focused on developing a navigation and manipulation planner that takes explicitly into account the presence of humans in the environment.
I-Navigation in Human Presence
Introduction
Robot navigation in the presence of humans raises new issues for motion planning and control since the humans safety and comfort must be taken explicitly into account. We claim that a human-aware motion planner must not only elaborate safe robot paths, but also plan good, socially acceptable and legible paths. Our aim is to build a planner that takes explicitly into account the human partner by reasoning about his accessibility, his vision field and potential shared motions. This planner is part of a human-aware motion-manipulation planning and control system that we aim to develop in order to achieve motion and manipulation tasks in a collaborative way with the human.
Some Results
Simulation Results
|
A scenario with 2 humans looking each other. The robot chooses to pass between them because this path is visible and safe to the both of the humans. |
|
A scenario with 2 humans looking at the same direction. Although nothing is changed on the robots trajectory, the robot chooses an other path which is visible and safe to both of the humans. |
|
A scenario with 2 humans talking. Although the robot can pass between these two, it has to approach from behind and pass to close to them and so cause surprise. That is why the robot chooses an other path, longer that direct path but safer in a physical and mental sense. |
|
The same scenario with an obstacle placed next to the human on the right. This obstacle block the right part of the humans' field of view. Although the previous trajectory is valid, the robot chooses an other path to avoid bursting into the field of view by appearing suddenly behind the obstacle. |
|
The robot and the human meet in a hall. The robot plan his trajectory so that it can pass next to the human without causing any discomfort. After passing next to the human, instead of turning left immediately, the robot continues to go straight and turn left after a while to assure not to cause any fear. |
|
A living room scenario. The robot wants to go next to the human. It avoids to hide if possible, otherwise it tries not to burst into the field of view. |
|
A video showing the reactivity of the planner. All previous concepts are taken into account. |
Real Robot Implementation Results
 |
 |
|
A video showing a robot equipped with a classic navigation planner. As it doesn't take into account humans explicitly, the resulting path are not comfortable. |
A video showing a robot equipped with our Human Aware navigation planner. The robot modifies his trajectory according to the humans' positions and their orientations. |
 |
 |
| A video showing a robot equipped with our Human Aware navigation planner. According to the human's orientation, as the robot is visible when passing in front of the human, it passes closer than when it's invisible. |
A video showing a robot equipped with our Human Aware navigation planner. According to the human's orientation, the robot calculates a path that keeps a larger distance to the human's back. |
 |
 |
| Human Robot crossing situation. The robot gets out of the human's way. After passing him, unlike classical motion planners it doesn't resume immediately its path but gives a certain distance from the human's back. |
A scenario where the robot is blocked by an obstacle from the human's view. Instead of bursting right behind the obstacle and causing surprise/fear to the human, it calculates a path in which the robot enters the f.o.v. from a farther point. |
and related papers
A Human Aware Mobile Robot Motion Planner,
Emrah Akin Sisbot, Luis F. Marin-Urias, Rachid Alami and Thierry Simeon,
IEEE Transactions on Robotics, Volume: 23, Issue: 5, pp: 874-883, October, 2007.
A mobile robot that performs human acceptable motion (paper),
E. Akin Sisbot, Luis F. Marin Urias, Rachid Alami and Thierry Simeon,
2006 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS2006), Beijing, China.
Implementing a Human-Aware Robot System (paper),
E. Akin Sisbot, Aurelie Clodic, Luis F. Marin Urias, Mathias Fontmarty, Ludovic Brthes, Rachid Alami,
IEEE International Symposium on Robot and Human Interactive Communication 2006 (RO-MAN 06), Hatfield, U.K.
How may I serve you? A robot companion approaching a seated person in a helping context (paper),
K. Dautenhahn, M. Walters, S. Woods, K. Lee Koay, E. A. Sisbot, R. Alami, T. Simeon,
HRI Human Robot Interaction '06 - HRI2006, Salt Lake City, Utah, USA.
Navigation in the Presence of Humans (paper)
E. A. Sisbot, R. Alami and T. Simeon, K. Dautenhahn, M. Walters, S. Woods, K. L. Koay and C. Nehaniv,
IEEE-RAS International Conference on Humanoid Robots Humanoids2005 December 5-7, 2005 Tsukuba Japan.
Architecture of the placement planner for human-robot close interaction,
E. A. Sisbot and R. Alami,
Cogniron Deliverable D3.5.1, LAAS Internal Report, January 2005
II-Manipulation in Human Presence
Introduction
Our approach is based on separating the whole problem of manipulation, e.g a robot giving an object to the human, into 3 stages. Each of these stages will produce the corresponding result and past it to the next stage:
Spatial coordinates of the point where the object will be handled to the human,
The path that the object will follow from its resting position to human hand as it was a free flying object,
The path of the whole body of the robot along with its posture for manipulation.
1- Calculating object position
One of the key points in the manipulation planning is to decide where robot, human and the object meet. In classical motion planners, this decision is made implicitly by only reasoning about robot's and the object's structure. The absence of human is compensated by letting him adapt himself to the robot's motion, thus making the duty of the human more important and the motions of the robot less predictable.We present 3 properties of the interaction, safety, visibility and arm comfort, that will help us to find safe and comfortable coordinates of the object where the robot will handle it to the human. These properties are represented as 3D grids and most suitable (lower cost) point in this grid is assigned to be the point where the robot will place the object for the human to handle.
2- Calculating object path
As we found where the robot must place the object in the previous stage, we now have to find the path that the object will take from its initial position to this final position. To find this path we use a 3D grid based approach which we build around the human.This grid contains a set of cells with various costs derived from the relative configuration of the human, his state, his capabilities and preferences. A minimum-cost path search from object's current position to the previously found object exchange position is performed and this path is assigned to be the object's path.
3- Calculating robot path
Eventhough we found a path for the object (and robot's hand) to follow, it is not enough to produce a acceptable robot motion in HRI context where the motion should be safe, comfortable and predictable. With this motion the robot must make clear of its intention.The third and final stage of planning consists of finding a path for the robot that will follow object's motion. The object's motion is computed as it was a freeflying object. But in reality it is the robot who holds the object and who will make the object follow it's path.
To adapt the robot structure to the object's motion, we use Generalized Inverse Kinematics algorithm with two tasks.
Some Results
Simulation Results
|
A video of robot handling object motion by using a classical motion planner. The motion is direct and does not take into account human's comfort. |
|
The previous scenario with a robot equipped with the "Human Aware Manipulation Planner". The planner produces a comfortable motion to handle the object to the human. The motion takes into account the visibility, the safety and the arm comfort of the human. It also take into account the fact that the human is right handed and sitting. Note that during its hand motion, the robot also looks to the object, thus better express its intention. |
|
In this scenario human is standing. The robot takes into account this change and recalculates a new path. |
|
Using Generalized Inverse Kinematics method allows us to easily change robots. The algorithms are implementable to different types of robot. Different sub tasks can be given according to the speciality of robot structure. This example illustrates the same scenario with a mobile manipulator robot. |
|
And also a humanoid robot working in the previous scenario equipped with Human Aware Manipulation Planner. |
Real Robot Implementation Results
 |
 |
|
A video showing a robot handling an object to a sitting person. The planner module uses leg detection data and visual hand/head detection data to construct a 3D model of the human. The planner generates a path towards the human hand once the person makes a handling motion. This path is then sent to the limited jerk arm execution module which results friendly and safe motions. |
The same scenario expect this time the person is standing. With the help of detection/tracking modules, the planner constructs a 3d sitting person in its environment. |
 |
A video showing a robot transfers a bottle from one lazy person to another. The planner generates a path towards the bottle hand. The presence of the bottle in robot's grippers is sensed by the force sensor. Once the robots has the bottle, it turns its cameras to model the second person. The path is generated once this person makes the handling motion and then sent to the limited jerk arm execution module which results friendly and safe motions. |
and related papers
Spatial Reasoning for Human Robot Interaction, (paper)
Emrah Akin Sisbot, Luis Felipe Marin and Rachid Alami,
2007 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2007), San Diego, USA.
Supervision and Motion Planning for a Mobile Manipulator Interacting with Humans (paper),
Emrah Akin Sisbot, Aurelie Clodic, Rachid Alami, Maxime Ransan,
2008 ACM/IEEE International Conference on Human-Robot Interaction (HRI 2008), Amsterdam, Netherlands.