Study on networked collaboration using haptic devices

由于近年来通信网络的发展，不仅可以传输语音和图像等传统媒体，还可以传输触觉信息等其他流媒体。这是一项非常具有挑战性的任务，并且已经提出了一些在网络上使用触觉信息的实验结果。由于几乎所有的研究都是由机器人研究人员社区进行的，他们“天真地”忽略了延迟、抖动和丢包。也就是说，他们认为网络是理想的。然而，现实网络与理想网络相去甚远，网络损伤导致触觉应用的服务质量（QoS）严重恶化。从这个角度出发，我们研究了网络损伤，特别是时延、抖动和丢包对网络触觉协同的影响。例如，我们发现100[毫秒]的延迟在触觉协作中是非常关键的，如果超过100[毫秒]。，我们无法以任何方式保持良好的QoS。然后，在此基础上，提出了应对此类损害的对策。针对丢包问题，提出了一种实用的航位推算方法。针对带宽的限制，提出了一种基于航位推算和二阶预测的带宽压缩方法。针对抖动，我们首先提出了一种实用的媒体内同步机制。我们还提出了一种跨媒体同步机制。最后，我们将上述对策整合为一个实际的系统。它具有很强的鲁棒性和令人满意的特点，作为网络触觉协同系统对抗网络损伤的实际手段，已得到广泛的应用。

Owing to recent advance of communication networks, it is now becoming realistic to transmit not only conventional media like voice and images but also other streaming media as haptic information. This is a very challenging task and a few experimental result of the usage of haptic information over networks have been proposed. Since almost all the researches were by the community of researchers on robotics, they "innocently" ignored the delay, jitter and packet loss. Namely, they treated the network as ideal. However, the real networks are far from ideal and network impairment causes the heavy deteriation of Quality of Services (QoS) on haptic application. From this viewpoint, we study the effect of network impairment, especially delay, jitter and packet loss, on the haptics collaboration on Internet. For example, we found that delay of 100 [msec] is very crucial in haptic collaboration and if it is more than 100 [msec.], we cannot maintain good QoS in any way. Then, based on them, several countermeasures against such impairment were proposed. Against packet loss, we proposed a practical dead-reckoning method. Against, the limitation of bandwidth, we proposed a bandwidth compression method based on dead-reckoning and the 2nd order prediction. Against jitter, we first proposed a practical intra-media synchronization mechanism. We also proposed an inter-media synchronization mechanism. Finally, we integrated the above mentioned countermeasure into one practical system. It is very and robust and happily, it is now becoming widely used as the de-fact of the countermeasure against network impairment on networked haptic collaboration systems.