There are two approaches to Trajectory planning, “On-line” and “Off-line” planning. Depending on environmental conditions any of these two could be applicable. Generally we classify robot environments into two static and dynamicenvironments. Environments withimmovable object and targets are called static environments, while the robot workspaces witch may not be static and may permenantly change in time are called dynamic environments.Dynamic environments can be divided into structured and un structured dynamic environments, depending on our knowledge of changes of the environment. One of the problems considered in the on-line trajectory planning for a robot in an unstructured dynamic environment is the problem of catching a moving target with unknown motion. This problem is studied extensively in this thesis for serial manipulators. Moving objects to be intercepted by a robot arm can beclassify as follows: Slow-maneuvering object, which moves on a continuous path at a relatively constant velocity or at constant acceleration;Fast-maneuvering object, which moves with an arbitrary random acceleration;Various methods are developed for these problems. An On-line point to point trajectory planning method based on adaptive prediction, planning and execution (APPE) trajectory planning is proposed for slow-maneuvering objects. In this method a primary trajectory is designed and will be repaired several times in the middle of robot motion considering environment changes.A time optimal trajectory planning is developed for this reason and is presented. For interception of a fast-maneuvering object, Navigation Guidance Methods are used. Four related method are reported for this approach. Proportional Navigation Guidance (PNG), Ideal PNG (IPNG), Augmented PNG (APNG) and Augmented Ideal PNG (AIPNG) are studied and there performance are analyzed and compared through numerical simulation for two planar and spatial serial manipulators. In this method an acceleration vector is forced to the robot end effectors aiming to close it to the target and catch it. Catching will not be smooth using these navigation laws. In order to obtain a soft interception, a modified AIPNG is proposed and its performance is compared with AIPNG. this approach was developed for 3D problems.Object condition must be known in every instance. Vision based control method are used for this reason.Finally an experimental setup is developed to test the proposed online trajectory planning methods. Object Condition is acquired throw a vision system and image processing algorithms. Test results for trajectory planning based on APPE method are presented. Keywords: Catching, Moving objects, Dynamic environment, On-line trajectory planning